diff --git a/docs/ml-conversational-analytic-tool/baseCNN.html b/docs/ml-conversational-analytic-tool/baseCNN.html index 4005539..a1b771b 100644 --- a/docs/ml-conversational-analytic-tool/baseCNN.html +++ b/docs/ml-conversational-analytic-tool/baseCNN.html @@ -35,18 +35,21 @@

Module ml-conversational-analytic-tool.baseCNNModule ml-conversational-analytic-tool.baseCNNModule ml-conversational-analytic-tool.baseCNNModule ml-conversational-analytic-tool.baseCNNModule ml-conversational-analytic-tool.baseCNN @@ -117,22 +133,26 @@

Classes

class BaseCNN
-
+

Constructor creates model and explainer

Expand source code 
class BaseCNN:
-    # Constructor creates model and explainer
     def __init__(self):
+        """
+        Constructor creates model and explainer
+        """
         self.dimension2 = True
         self.input_shape = ()
         self.model = keras.models.Sequential()
         self.explainer = SmoothGrad()
         self.model_ready = False
 
-    # Make 1d model for role agnostic data
     def makeModel(self, input_shape):
+        """
+        Make 1d model for role agnostic data
+        """
         self.input_shape = input_shape
         self.model.add(keras.layers.Conv1D(32, 3, activation='relu', input_shape=input_shape))
         self.model.add(keras.layers.MaxPooling1D(2))
@@ -143,8 +163,10 @@ 
Classes

         self.dimension2 = False
         self.model_ready = True
 
-    # Make 2d model for role relevant data
     def makeModel2D(self, input_shape):
+        """
+        Make 2d model for role relevant data
+        """
         self.input_shape = input_shape
         self.model.add(keras.layers.Conv2D(4, (5, 5), activation='relu', input_shape=input_shape))
         self.model.add(keras.layers.MaxPooling2D((4, 4)))
@@ -156,8 +178,10 @@ 
Classes

         self.dimension2 = True
         self.model_ready = True
 
-    # Train model
     def trainModel(self, obs, res, val_split=0.3, val_set=None, epochs=10, batch_size=32):
+        """
+        Train model
+        """
         self.model.compile(optimizer=keras.optimizers.Adam(), loss='binary_crossentropy', metrics=['accuracy'])
         if val_set:
             train_hist = self.model.fit(np.array(obs), np.array(res), epochs=epochs, batch_size=batch_size,
@@ -168,8 +192,13 @@ 
Classes

                                         validation_split=val_split, verbose=1)
             return train_hist
 
-    # Score model for accuracy, precision and recall
+    def saveModel(self, name, version):
+        self.model.save("{}/{}".format(name, version))
+
     def scoreModel(self, obs, res):
+        """
+        Score model for accuracy, precision and recall
+        """
         evaluation = {}
         evaluation['Loss_Acc'] = self.model.evaluate(np.array(obs), np.array(res))
         evaluation['Precision_Recall_Fscore_Support'] = precision_recall_fscore_support(res, self.predict(obs, True),
@@ -177,15 +206,19 @@ 
Classes

         print("Accuracy: {}".format(evaluation['Loss_Acc'][1]))
         return evaluation
 
-    # Get predictions
     def predict(self, obs, labels=False):
+        """
+        Get predictions
+        """
         predictions = self.model.predict(np.array(obs))
         if labels:
             return [1 if x > 0.5 else 0 for x in predictions]
         return predictions
 
-    # Explain prediction for obs using explainer
     def explain(self, obs):
+        """
+        Explain prediction for obs using explainer
+        """
         output = self.explainer.explain((obs, None), self.model, 1, 20, 1.)
         return output
@@ -195,12 +228,15 @@

Methods

def explain(self, obs)
-
+

Explain prediction for obs using explainer

Expand source code 
def explain(self, obs):
+    """
+    Explain prediction for obs using explainer
+    """
     output = self.explainer.explain((obs, None), self.model, 1, 20, 1.)
     return output
@@ -209,12 +245,15 @@

Methods

def makeModel(self, input_shape)
-
+

Make 1d model for role agnostic data

Expand source code 
def makeModel(self, input_shape):
+    """
+    Make 1d model for role agnostic data
+    """
     self.input_shape = input_shape
     self.model.add(keras.layers.Conv1D(32, 3, activation='relu', input_shape=input_shape))
     self.model.add(keras.layers.MaxPooling1D(2))
@@ -230,12 +269,15 @@ 
Methods

 def makeModel2D(self, input_shape)
 
 

-

+
Make 2d model for role relevant data

 

 
 Expand source code
 
 
def makeModel2D(self, input_shape):
+    """
+    Make 2d model for role relevant data
+    """
     self.input_shape = input_shape
     self.model.add(keras.layers.Conv2D(4, (5, 5), activation='relu', input_shape=input_shape))
     self.model.add(keras.layers.MaxPooling2D((4, 4)))
@@ -252,28 +294,47 @@ 
Methods

 def predict(self, obs, labels=False)
 
 

-

+
Get predictions

 

 
 Expand source code
 
 
def predict(self, obs, labels=False):
+    """
+    Get predictions
+    """
     predictions = self.model.predict(np.array(obs))
     if labels:
         return [1 if x > 0.5 else 0 for x in predictions]
     return predictions

 

 

+

+def saveModel(self, name, version)
+

+

+

+

+
+Expand source code
+
+
def saveModel(self, name, version):
+    self.model.save("{}/{}".format(name, version))

+

+

 

 def scoreModel(self, obs, res)
 

 

-

+
Score model for accuracy, precision and recall

 

 
 Expand source code
 
 
def scoreModel(self, obs, res):
+    """
+    Score model for accuracy, precision and recall
+    """
     evaluation = {}
     evaluation['Loss_Acc'] = self.model.evaluate(np.array(obs), np.array(res))
     evaluation['Precision_Recall_Fscore_Support'] = precision_recall_fscore_support(res, self.predict(obs, True),
@@ -286,12 +347,15 @@ 
Methods

 def trainModel(self, obs, res, val_split=0.3, val_set=None, epochs=10, batch_size=32)
 
 

-

+
Train model

 

 
 Expand source code
 
 
def trainModel(self, obs, res, val_split=0.3, val_set=None, epochs=10, batch_size=32):
+    """
+    Train model
+    """
     self.model.compile(optimizer=keras.optimizers.Adam(), loss='binary_crossentropy', metrics=['accuracy'])
     if val_set:
         train_hist = self.model.fit(np.array(obs), np.array(res), epochs=epochs, batch_size=batch_size,
@@ -328,6 +392,7 @@ 
makeModel
 
  • makeModel2D
    • 
 
  • predict
    • 
+
  • saveModel
    • 
 
  • scoreModel
    • 
 
  • trainModel
    • 
 
diff --git a/docs/ml-conversational-analytic-tool/baseLSTM.html b/docs/ml-conversational-analytic-tool/baseLSTM.html
index c11717a..e8fc887 100644
--- a/docs/ml-conversational-analytic-tool/baseLSTM.html
+++ b/docs/ml-conversational-analytic-tool/baseLSTM.html
@@ -35,17 +35,20 @@ 
    Module ml-conversational-analytic-tool.baseLSTMModule ml-conversational-analytic-tool.baseLSTMModule ml-conversational-analytic-tool.baseLSTMModule ml-conversational-analytic-tool.baseLSTMModule ml-conversational-analytic-tool.baseLSTMClasses
    
 class BaseLSTM
     
 
    
-
    
+
    Constructor creates model
    
 
    
 
 Expand source code
 
 
    class BaseLSTM:
-    # Constructor creates model
     def __init__(self):
+        """
+        Constructor creates model
+        """
         self.dimension2 = True
         self.input_shape = ()
         self.model = None
         self.model_ready = False
 
-    # Make lstm model for role agnostic data
     def makeModel(self, input_shape):
+        """
+        Make lstm model for role agnostic data
+        """
         self.input_shape = input_shape
         self.model = keras.models.Sequential()
         self.model.add(keras.layers.LSTM(8, input_shape=input_shape))
@@ -164,22 +185,29 @@ 
    Classes
    
         self.dimension2 = False
         self.model_ready = True
 
-    # Made lstm model for role relevant layers data
     def makeModel2D(self, input_shape):
+        """
+        Made lstm model for role relevant layers data
+        """
+
         # Inputs
         self.input_shape = input_shape
         inputs = keras.Input(shape=(None, 512, 2), dtype="float32")
+
         # Pipe output of author and reviewer layer to two lstm
         author = keras.layers.Lambda(lambda x: x[:, :, :, 0])(inputs)
         reviewer = keras.layers.Lambda(lambda x: x[:, :, :, 1])(inputs)
+
         # Create author lstm
         author_lstm = keras.layers.LSTM(4, activation='relu', return_sequences=False)(author)
         author_dropout = keras.layers.Dropout(0.2)(author_lstm)
         author_output = keras.layers.Dense(1, activation='relu')(author_dropout)
+
         # Create reviewer lstm
         reviewer_lstm = keras.layers.LSTM(4, activation='relu', return_sequences=False)(reviewer)
         reviewer_dropout = keras.layers.Dropout(0.2)(reviewer_lstm)
         reviewer_output = keras.layers.Dense(1, activation='relu')(reviewer_dropout)
+
         # Concatenate author and reviewer output
         combine = keras.layers.Concatenate(axis=1)([author_output, reviewer_output])
         output = keras.layers.Dense(1, activation='sigmoid')(combine)
@@ -189,7 +217,6 @@ 
    Classes
    
         self.dimension2 = True
         self.model_ready = True
 
-    # Train model
     def trainModel(self, obs, res, val_split=0.3, val_set=None, epochs=10, batch_size=32):
         self.model.compile(optimizer=keras.optimizers.Adam(), loss='binary_crossentropy', metrics=['accuracy'])
         if val_set:
@@ -203,8 +230,13 @@ 
    Classes
    
                                         batch_size=batch_size, verbose=1)
             return train_hist
 
-    # Score model for accuracy, precision and recall
+    def saveModel(self, name, version):
+        self.model.save("{}/{}".format(name, version))
+
     def scoreModel(self, obs, res):
+        """
+        Score model for accuracy, precision and recall
+        """
         evaluation = {}
         evaluation['Loss_Acc'] = self.model.evaluate(tf.ragged.stack(obs), tf.convert_to_tensor(res))
         evaluation['Precision_Recall_Fscore_Support'] = precision_recall_fscore_support(res, self.predict(obs, True),
@@ -212,21 +244,24 @@ 
    Classes
    
         print("Accuracy: {}".format(evaluation['Loss_Acc'][1]))
         return evaluation
 
-    # Get predictions
     def predict(self, obs, labels=False):
         predictions = self.model.predict(tf.ragged.stack(obs))
         if labels:
             return [1 if x > 0.5 else 0 for x in predictions]
         return predictions
 
-    # Explain prediction for obs by calculating gradients
     def explain(self, obs):
+        """
+        Explain prediction for obs by calculating gradients
+        """
         grads = self._gradientImportance(obs)
         imp = grads[:, 0] + grads[:, 1]
         return imp
 
-    # Get gradients
     def _gradientImportance(self, seq):
+        """
+        Get gradients
+        """
         seq = tf.Variable(seq[np.newaxis, :, :])
         with tf.GradientTape() as tape:
             predictions = self.model(seq)
@@ -241,12 +276,15 @@ 
    Methods
    
 def explain(self, obs)
     
 
    
-
    
+
    Explain prediction for obs by calculating gradients
    
 
    
 
 Expand source code
 
 
    def explain(self, obs):
+    """
+    Explain prediction for obs by calculating gradients
+    """
     grads = self._gradientImportance(obs)
     imp = grads[:, 0] + grads[:, 1]
     return imp
    
@@ -256,12 +294,15 @@ 
    Methods
    
 def makeModel(self, input_shape)
 
 
    
-
    
+
    Make lstm model for role agnostic data
    
 
    
 
 Expand source code
 
 
    def makeModel(self, input_shape):
+    """
+    Make lstm model for role agnostic data
+    """
     self.input_shape = input_shape
     self.model = keras.models.Sequential()
     self.model.add(keras.layers.LSTM(8, input_shape=input_shape))
@@ -276,26 +317,34 @@ 
    Methods
    
 def makeModel2D(self, input_shape)
     
 
    
-
    
+
    Made lstm model for role relevant layers data
    
 
    
 
 Expand source code
 
 
    def makeModel2D(self, input_shape):
+    """
+    Made lstm model for role relevant layers data
+    """
+
     # Inputs
     self.input_shape = input_shape
     inputs = keras.Input(shape=(None, 512, 2), dtype="float32")
+
     # Pipe output of author and reviewer layer to two lstm
     author = keras.layers.Lambda(lambda x: x[:, :, :, 0])(inputs)
     reviewer = keras.layers.Lambda(lambda x: x[:, :, :, 1])(inputs)
+
     # Create author lstm
     author_lstm = keras.layers.LSTM(4, activation='relu', return_sequences=False)(author)
     author_dropout = keras.layers.Dropout(0.2)(author_lstm)
     author_output = keras.layers.Dense(1, activation='relu')(author_dropout)
+
     # Create reviewer lstm
     reviewer_lstm = keras.layers.LSTM(4, activation='relu', return_sequences=False)(reviewer)
     reviewer_dropout = keras.layers.Dropout(0.2)(reviewer_lstm)
     reviewer_output = keras.layers.Dense(1, activation='relu')(reviewer_dropout)
+
     # Concatenate author and reviewer output
     combine = keras.layers.Concatenate(axis=1)([author_output, reviewer_output])
     output = keras.layers.Dense(1, activation='sigmoid')(combine)
@@ -322,16 +371,32 @@ 
    Methods
    
     return predictions
    
 
    
 
    
+
    
+def saveModel(self, name, version)
+
    
+
    
+
    
+
    
+
+Expand source code
+
+
    def saveModel(self, name, version):
+    self.model.save("{}/{}".format(name, version))
    
+
    
+
    
 
    
 def scoreModel(self, obs, res)
 
    
 
    
-
    
+
    Score model for accuracy, precision and recall
    
 
    
 
 Expand source code
 
 
    def scoreModel(self, obs, res):
+    """
+    Score model for accuracy, precision and recall
+    """
     evaluation = {}
     evaluation['Loss_Acc'] = self.model.evaluate(tf.ragged.stack(obs), tf.convert_to_tensor(res))
     evaluation['Precision_Recall_Fscore_Support'] = precision_recall_fscore_support(res, self.predict(obs, True),
@@ -388,6 +453,7 @@ 
    makeModel
 
  • makeModel2D
    • 
 
  • predict
    • 
+
  • saveModel
    • 
 
  • scoreModel
    • 
 
  • trainModel
    • 
 
diff --git a/docs/ml-conversational-analytic-tool/commentAnalysis.html b/docs/ml-conversational-analytic-tool/commentAnalysis.html
index 70caebc..e98c683 100644
--- a/docs/ml-conversational-analytic-tool/commentAnalysis.html
+++ b/docs/ml-conversational-analytic-tool/commentAnalysis.html
@@ -37,17 +37,21 @@ 
    Module ml-conversational-analytic-tool.commentAnalysisModule ml-conversational-analytic-tool.commentAnalysisModule ml-conversational-analytic-tool.commentAnalysisClasses
    
 (words)
 
 
    
-
    
+
    Constructors form a dictionary to be used for counting.
+Parameters: words - list of words to count
    
 
    
 
 Expand source code
 
 
    class CommentAnalyzer:
-    # Constructors form a dictionary to be used for counting.
-    # Parameters: words - list of words to count
     def __init__(self, words):
+        """
+        Constructors form a dictionary to be used for counting.
+        Parameters: words - list of words to count
+        """
         self.word_count = {word.lower(): 0 for word in words}  # Create dictionary with list items as key
         self.vader_sentiment = SentimentIntensityAnalyzer()  # Initialize sentiment analysis model
 
-    # Method to get desired features from an input comment.
-    # Parameters: comment - string. Returns: dictionary with features
     def analyzeComment(self, comment):
+        """
+        Method to get desired features from an input comment.
+        Parameters: comment - string.
+        Returns: dictionary with features
+        """
         result = {}  # Create return dictionary
         cleaned_comment = self.preProcess(comment)  # Clean comment text
         # result['Word Counts'] = self.countWords(cleaned_comment)  # Determine word counts
@@ -143,16 +166,22 @@ 
    Classes
    
         result['Code Blocks'] = self.getCodeBlockCount(cleaned_comment)  # Determine code block count
         return result
 
-    # Method to clean and return text. Parameters: text - string.
-    # Returns: string after cleaning
     def preProcess(self, text):
+        """
+        Method to clean and return text.
+        Parameters: text - string.
+        Returns: string after cleaning
+        """
         cleaned_text = text.strip()  # Remove trailing and starting spaces
         cleaned_text = cleaned_text.lower()  # Convert to lowercase
         return cleaned_text
 
-    # Method to determine word count. Parameters: comment - string
-    # Returns: dictionary with word counts
     def countWords(self, comment):
+        """
+        Method to determine word count.
+        Parameters: comment - string
+        Returns: dictionary with word counts
+        """
         words = comment.split(" ")  # Split text into words
         current_word_count = self.word_count.copy()  # Copy default dict for new count
         for word in words:  # Iterate over all words
@@ -160,22 +189,29 @@ 
    Classes
    
                 current_word_count[word] = current_word_count[word] + 1
         return current_word_count
 
-    # Method to determine the code blocks. Parameters: comment - string
-    # Returns: integer count
     def getCodeBlockCount(self, comment):
+        """
+        Method to determine the code blocks.
+        Parameters: comment - string
+        Returns: integer count
+        """
         count = comment.count("```")  # Find occurences of code block
         if count % 2 != 0:  # Should be in pairs
             print("Warning: Mismatched code blocks")
             return int(count / 2 - 1)  # Subtract 1 since unmatched pair
         return int(count / 2)  # Divide by 2 since pairs
 
-    # Method to determine sentiment. Parameters: comment - string
-    # Returns: dictionary with positive, negative and neutral scores
     def getSentiment(self, comment):
+        """
+        Method to determine sentiment. Parameters: comment - string
+        Returns: dictionary with positive, negative and neutral scores
+        """
         return self.vader_sentiment.polarity_scores(comment)["compound"]
 
-    # Method to change words to count. Parameters: Set new word count with new keys/
     def changeWords(self, words):
+        """
+        Method to change words to count. Parameters: Set new word count with new keys/
+        """
         self.word_count = {word: 0 for word in words}
    
 
    
 
    Methods
    
@@ -184,12 +220,19 @@ 
    Methods
    
 def analyzeComment(self, comment)
     
 
    
-
    
+
    Method to get desired features from an input comment.
+Parameters: comment - string.
+Returns: dictionary with features
    
 
    
 
 Expand source code
 
 
    def analyzeComment(self, comment):
+    """
+    Method to get desired features from an input comment.
+    Parameters: comment - string.
+    Returns: dictionary with features
+    """
     result = {}  # Create return dictionary
     cleaned_comment = self.preProcess(comment)  # Clean comment text
     # result['Word Counts'] = self.countWords(cleaned_comment)  # Determine word counts
@@ -203,12 +246,15 @@ 
    Methods
    
 def changeWords(self, words)
     
 
    
-
    
+
    Method to change words to count. Parameters: Set new word count with new keys/
    
 
    
 
 Expand source code
 
 
    def changeWords(self, words):
+    """
+    Method to change words to count. Parameters: Set new word count with new keys/
+    """
     self.word_count = {word: 0 for word in words}
    
 
    
 
    
@@ -216,12 +262,19 @@ 
    Methods
    
 def countWords(self, comment)
     
 
    
-
    
+
    Method to determine word count.
+Parameters: comment - string
+Returns: dictionary with word counts
    
 
    
 
 Expand source code
 
 
    def countWords(self, comment):
+    """
+    Method to determine word count.
+    Parameters: comment - string
+    Returns: dictionary with word counts
+    """
     words = comment.split(" ")  # Split text into words
     current_word_count = self.word_count.copy()  # Copy default dict for new count
     for word in words:  # Iterate over all words
@@ -234,12 +287,19 @@ 
    Methods
    
 def getCodeBlockCount(self, comment)
     
 
    
-
    
+
    Method to determine the code blocks.
+Parameters: comment - string
+Returns: integer count
    
 
    
 
 Expand source code
 
 
    def getCodeBlockCount(self, comment):
+    """
+    Method to determine the code blocks.
+    Parameters: comment - string
+    Returns: integer count
+    """
     count = comment.count("```")  # Find occurences of code block
     if count % 2 != 0:  # Should be in pairs
         print("Warning: Mismatched code blocks")
@@ -251,12 +311,17 @@ 
    Methods
    
 def getSentiment(self, comment)
     
 
    
-
    
+
    Method to determine sentiment. Parameters: comment - string
+Returns: dictionary with positive, negative and neutral scores
    
 
    
 
 Expand source code
 
 
    def getSentiment(self, comment):
+    """
+    Method to determine sentiment. Parameters: comment - string
+    Returns: dictionary with positive, negative and neutral scores
+    """
     return self.vader_sentiment.polarity_scores(comment)["compound"]
    
 
    
 
    
@@ -264,12 +329,19 @@ 
    Methods
    
 def preProcess(self, text)
 
 
    
-
    
+
    Method to clean and return text.
+Parameters: text - string.
+Returns: string after cleaning
    
 
    
 
 Expand source code
 
 
    def preProcess(self, text):
+    """
+    Method to clean and return text.
+    Parameters: text - string.
+    Returns: string after cleaning
+    """
     cleaned_text = text.strip()  # Remove trailing and starting spaces
     cleaned_text = cleaned_text.lower()  # Convert to lowercase
     return cleaned_text
    
diff --git a/docs/ml-conversational-analytic-tool/featureVector.html b/docs/ml-conversational-analytic-tool/featureVector.html
index 0e2b581..6fbfe77 100644
--- a/docs/ml-conversational-analytic-tool/featureVector.html
+++ b/docs/ml-conversational-analytic-tool/featureVector.html
@@ -38,12 +38,12 @@ 
    Module ml-conversational-analytic-tool.featureVectorModule ml-conversational-analytic-tool.featureVectorModule ml-conversational-analytic-tool.featureVectorModule ml-conversational-analytic-tool.featureVectorModule ml-conversational-analytic-tool.featureVectorModule ml-conversational-analytic-tool.featureVectorClasses
    
 (retain_features, analysis_features)
 
 
    
-
    
+
    Constructor sets instance variables
+Inputs: Retained Features (list) -> pull data features, Analysis Features (list) - > comment analysis features
    
 
    
 
 Expand source code
 
 
    class Featurizer:
-    # Constructor sets instance variables
-    # Inputs: Retained Features (list) -> pull data features, Analysis Features (list) - > comment analysis features
     def __init__(self, retain_features, analysis_features):
+        """
+        Constructor sets instance variables
+        Inputs: Retained Features (list) -> pull data features, Analysis Features (list) - > comment analysis features
+        """
         self.analysis_features = analysis_features
         self.retain_pull_features = retain_features
         self.raw_filename = ""
@@ -266,9 +281,11 @@ 
    Classes
    
         self.featurized_data = None
         self.commentAnalyzer = None
 
-    # Function to read raw data stored as csv
-    # Inputs: File location (string) -> input raw data file location
     def readRawData(self, filename):
+        """
+        Function to read raw data stored as csv
+        Inputs: File location (string) -> input raw data file location
+        """
         self.raw_filename = filename
         self.raw_data = pd.read_csv(filename)
 
@@ -284,9 +301,11 @@ 
    Classes
    
         self.raw_data['Review_Comments'] = self.raw_data['Review_Comments'].apply(lambda comment: stringToDict(comment))
         print("Done reading raw data.")
 
-    # Function to obtain Comment Analyzer with parameters
-    # Inputs: File location (string) -> .txt containing keywords to count
     def setupCommentAnalyzer(self, filename):
+        """
+        Function to obtain Comment Analyzer with parameters
+        Inputs: File location (string) -> .txt containing keywords to count
+        """
         word_list = []
 
         # Open file to obtain list of words in Comment Analzyer
@@ -296,9 +315,11 @@ 
    Classes
    
         self.analysis_features = self.analysis_features + word_list
         print("Comment Analyzer Setup")
 
-    # Function to create/export dataset with desired features
-    # Inputs: Name of file to export (string), export flag (boolean)
     def formFeatures(self, export_filename="", export=True):
+        """
+        Function to create/export dataset with desired features
+        Inputs: Name of file to export (string), export flag (boolean)
+        """
         features = []  # List of rows to convert to dataframe
 
         # Iterate over each pull
@@ -391,12 +412,16 @@ 
    Classes
    
 
         return export_df
 
-    # Function to form conversation string given communication on a pull request
-    # Inputs: Name of file to export (string), export flag (boolean)
     def pullStringConversation(self, export_filename="", export=True):
+        """
+        Function to form conversation string given communication on a pull request
+        Inputs: Name of file to export (string), export flag (boolean)
+        """
+
         # Store each interaction and pull URL for export
         string_conversations = []
         pull_urls = []
+        pull_numbers = []
 
         for index, row in self.raw_data.iterrows():
             # Make pull message
@@ -415,9 +440,11 @@ 
    Classes
    
                                                                           comment_row["Body"])
             string_conversations.append(conversation.encode("ascii", "ignore").decode())
             pull_urls.append(row["URL"])
+            pull_numbers.append(row["Number"])
 
         # Export converation field dataset
         export_df = pd.DataFrame()
+        export_df["Number"] = pull_numbers
         export_df["URL"] = pull_urls
         export_df["Thread"] = string_conversations
 
@@ -436,12 +463,17 @@ 
    Methods
    
 def formFeatures(self, export_filename='', export=True)
     
 
    
-
    
+
    Function to create/export dataset with desired features
+Inputs: Name of file to export (string), export flag (boolean)
    
 
    
 
 Expand source code
 
 
    def formFeatures(self, export_filename="", export=True):
+    """
+    Function to create/export dataset with desired features
+    Inputs: Name of file to export (string), export flag (boolean)
+    """
     features = []  # List of rows to convert to dataframe
 
     # Iterate over each pull
@@ -539,15 +571,22 @@ 
    Methods
    
 def pullStringConversation(self, export_filename='', export=True)
     
 
    
-
    
+
    Function to form conversation string given communication on a pull request
+Inputs: Name of file to export (string), export flag (boolean)
    
 
    
 
 Expand source code
 
 
    def pullStringConversation(self, export_filename="", export=True):
+    """
+    Function to form conversation string given communication on a pull request
+    Inputs: Name of file to export (string), export flag (boolean)
+    """
+
     # Store each interaction and pull URL for export
     string_conversations = []
     pull_urls = []
+    pull_numbers = []
 
     for index, row in self.raw_data.iterrows():
         # Make pull message
@@ -566,9 +605,11 @@ 
    Methods
    
                                                                       comment_row["Body"])
         string_conversations.append(conversation.encode("ascii", "ignore").decode())
         pull_urls.append(row["URL"])
+        pull_numbers.append(row["Number"])
 
     # Export converation field dataset
     export_df = pd.DataFrame()
+    export_df["Number"] = pull_numbers
     export_df["URL"] = pull_urls
     export_df["Thread"] = string_conversations
 
@@ -586,12 +627,17 @@ 
    Methods
    
 def readRawData(self, filename)
     
 
    
-
    
+
    Function to read raw data stored as csv
+Inputs: File location (string) -> input raw data file location
    
 
    
 
 Expand source code
 
 
    def readRawData(self, filename):
+    """
+    Function to read raw data stored as csv
+    Inputs: File location (string) -> input raw data file location
+    """
     self.raw_filename = filename
     self.raw_data = pd.read_csv(filename)
 
@@ -612,12 +658,17 @@ 
    Methods
    
 def setupCommentAnalyzer(self, filename)
     
 
    
-
    
+
    Function to obtain Comment Analyzer with parameters
+Inputs: File location (string) -> .txt containing keywords to count
    
 
    
 
 Expand source code
 
 
    def setupCommentAnalyzer(self, filename):
+    """
+    Function to obtain Comment Analyzer with parameters
+    Inputs: File location (string) -> .txt containing keywords to count
+    """
     word_list = []
 
     # Open file to obtain list of words in Comment Analzyer
diff --git a/docs/ml-conversational-analytic-tool/githubDataExtraction.html b/docs/ml-conversational-analytic-tool/githubDataExtraction.html
index fa70a3e..93adbb9 100644
--- a/docs/ml-conversational-analytic-tool/githubDataExtraction.html
+++ b/docs/ml-conversational-analytic-tool/githubDataExtraction.html
@@ -29,20 +29,20 @@ 
    Module ml-conversational-analytic-tool.githubDataExtract
 
    # Copyright 2021 VMware, Inc.
 # SPDX-License-Identifier: Apache-2.0
 
-import argparse
 import os
-# Imports
 import time
 
+import argparse
 import pandas as pd
 from github import Github
 from github.GithubException import RateLimitExceededException
 
 
-# Class to extract pull data from a github repo. The data includes reviews, comments, and other metadata from each pull.
 class GithubDataExtractor:
-    # Constructor requires an access token to start a Github session, and specifies instance variables
     def __init__(self, access_token):
+        """
+        Constructor requires an access token to start a Github session, and specifies instance variables
+        """
         self.g_ses = Github(access_token)  # Github object is used as a channel to the Github API
         self.current_repo = None  # Current Opended Repo
         self.reaction_flag = False
@@ -50,18 +50,22 @@ 
    Module ml-conversational-analytic-tool.githubDataExtract
         self.repo_name = ""
         self.organization = ""
 
-    # Method to  open (access) repository with given organization and repository name (reponame).
-    # Parameters: usernmae - owner of the repository, repo_name - name of repo to be opened
     def openRepo(self, organization, repo_name):
+        """
+        Method to  open (access) repository with given organization and repository name (reponame).
+        Parameters: username - owner of the repository, repo_name - name of repo to be opened
+        """
         self.current_repo = self.g_ses.get_repo(organization + "/" + repo_name)  # Open repo
         self.repo_opened = True
         self.repo_name = repo_name
         self.organization = organization
         print("Opened repo {} - {}".format(repo_name, organization))
 
-    # Method to form a dataframe containing pull information. Parameters: name - name of exported csv file,
-    # export - if the dataframe should be exported to csv. Returns: Dataframe with pull data
     def getAllPulls(self, name="", reaction_flag=False, export_to_csv=True):
+        """
+        Method to form a dataframe containing pull information. Parameters: name - name of exported csv file,
+        export - if the dataframe should be exported to csv. Returns: Dataframe with pull data
+        """
         self.reaction_flag = reaction_flag
         if self.repo_opened:  # Verify if a repo has been opened
             pull_data = []
@@ -78,9 +82,11 @@ 
    Module ml-conversational-analytic-tool.githubDataExtract
             return pull_df
         print("Please open a Repo")
 
-    # Extract pulls with given state. Parameters: state - state of the pull (open or closed)
-    # Return: list of dictionaries containing data regardining each pull
     def getPullsByState(self, state):
+        """
+        Extract pulls with given state. Parameters: state - state of the pull (open or closed)
+        Return: list of dictionaries containing data regardining each pull
+        """
         pull_data = []
         try:  # Call the Github api to get all pulls
             pulls = self.current_repo.get_pulls(state=state, sort='create')
@@ -98,9 +104,11 @@ 
    Module ml-conversational-analytic-tool.githubDataExtract
                 pull_data.append(self.getPullFeatures(pull))
         return pull_data
 
-    # Method to form a list of json strings rerpesenting comments (reviews or issue).
-    # Parameters: comments - list of comment objects. Returns: List of json strings
     def listOfComments(self, comments):
+        """
+        Method to form a list of json strings rerpesenting comments (reviews or issue).
+        Parameters: comments - list of comment objects. Returns: List of json strings
+        """
         list_comments = []
 
         # Iterate over each comment
@@ -142,9 +150,11 @@ 
    Module ml-conversational-analytic-tool.githubDataExtract
                                               "Updated_At": str(comment.updated_at)}))
         return list_comments
 
-    # Method to get all data for a particular pull. Parameters: pull - object representing a pull
-    # Returns: dictionary containing all data of a pull
     def getPullFeatures(self, pull):
+        """
+        Method to get all data for a particular pull. Parameters: pull - object representing a pull
+        Returns: dictionary containing all data of a pull
+        """
         pull_dict = {}
         pull_dict["Number"] = pull.number
         pull_dict["Title"] = pull.title
@@ -187,6 +197,7 @@ 
    Module ml-conversational-analytic-tool.githubDataExtract
     ACCESS_TOKEN = os.environ["GITACCESS"]  # Access Github token from environment for security purposes
     extractor = GithubDataExtractor(ACCESS_TOKEN)  # Create object
     extractor.openRepo(args.organization, args.reponame)  # Open repo
+
     # Extract all pulls and export them to .csv
     if args.filename:
         extractor.getAllPulls(args.filename, args.reactions)
@@ -208,14 +219,16 @@ 
    Classes
    
 (access_token)
     
 
    
-
    
+
    Constructor requires an access token to start a Github session, and specifies instance variables
    
 
    
 
 Expand source code
 
 
    class GithubDataExtractor:
-    # Constructor requires an access token to start a Github session, and specifies instance variables
     def __init__(self, access_token):
+        """
+        Constructor requires an access token to start a Github session, and specifies instance variables
+        """
         self.g_ses = Github(access_token)  # Github object is used as a channel to the Github API
         self.current_repo = None  # Current Opended Repo
         self.reaction_flag = False
@@ -223,18 +236,22 @@ 
    Classes
    
         self.repo_name = ""
         self.organization = ""
 
-    # Method to  open (access) repository with given organization and repository name (reponame).
-    # Parameters: usernmae - owner of the repository, repo_name - name of repo to be opened
     def openRepo(self, organization, repo_name):
+        """
+        Method to  open (access) repository with given organization and repository name (reponame).
+        Parameters: username - owner of the repository, repo_name - name of repo to be opened
+        """
         self.current_repo = self.g_ses.get_repo(organization + "/" + repo_name)  # Open repo
         self.repo_opened = True
         self.repo_name = repo_name
         self.organization = organization
         print("Opened repo {} - {}".format(repo_name, organization))
 
-    # Method to form a dataframe containing pull information. Parameters: name - name of exported csv file,
-    # export - if the dataframe should be exported to csv. Returns: Dataframe with pull data
     def getAllPulls(self, name="", reaction_flag=False, export_to_csv=True):
+        """
+        Method to form a dataframe containing pull information. Parameters: name - name of exported csv file,
+        export - if the dataframe should be exported to csv. Returns: Dataframe with pull data
+        """
         self.reaction_flag = reaction_flag
         if self.repo_opened:  # Verify if a repo has been opened
             pull_data = []
@@ -251,9 +268,11 @@ 
    Classes
    
             return pull_df
         print("Please open a Repo")
 
-    # Extract pulls with given state. Parameters: state - state of the pull (open or closed)
-    # Return: list of dictionaries containing data regardining each pull
     def getPullsByState(self, state):
+        """
+        Extract pulls with given state. Parameters: state - state of the pull (open or closed)
+        Return: list of dictionaries containing data regardining each pull
+        """
         pull_data = []
         try:  # Call the Github api to get all pulls
             pulls = self.current_repo.get_pulls(state=state, sort='create')
@@ -271,9 +290,11 @@ 
    Classes
    
                 pull_data.append(self.getPullFeatures(pull))
         return pull_data
 
-    # Method to form a list of json strings rerpesenting comments (reviews or issue).
-    # Parameters: comments - list of comment objects. Returns: List of json strings
     def listOfComments(self, comments):
+        """
+        Method to form a list of json strings rerpesenting comments (reviews or issue).
+        Parameters: comments - list of comment objects. Returns: List of json strings
+        """
         list_comments = []
 
         # Iterate over each comment
@@ -315,9 +336,11 @@ 
    Classes
    
                                               "Updated_At": str(comment.updated_at)}))
         return list_comments
 
-    # Method to get all data for a particular pull. Parameters: pull - object representing a pull
-    # Returns: dictionary containing all data of a pull
     def getPullFeatures(self, pull):
+        """
+        Method to get all data for a particular pull. Parameters: pull - object representing a pull
+        Returns: dictionary containing all data of a pull
+        """
         pull_dict = {}
         pull_dict["Number"] = pull.number
         pull_dict["Title"] = pull.title
@@ -353,12 +376,17 @@ 
    Methods
    
 def getAllPulls(self, name='', reaction_flag=False, export_to_csv=True)
     
 
    
-
    
+
    Method to form a dataframe containing pull information. Parameters: name - name of exported csv file,
+export - if the dataframe should be exported to csv. Returns: Dataframe with pull data
    
 
    
 
 Expand source code
 
 
    def getAllPulls(self, name="", reaction_flag=False, export_to_csv=True):
+    """
+    Method to form a dataframe containing pull information. Parameters: name - name of exported csv file,
+    export - if the dataframe should be exported to csv. Returns: Dataframe with pull data
+    """
     self.reaction_flag = reaction_flag
     if self.repo_opened:  # Verify if a repo has been opened
         pull_data = []
@@ -380,12 +408,17 @@ 
    Methods
    
 def getPullFeatures(self, pull)
     
 
    
-
    
+
    Method to get all data for a particular pull. Parameters: pull - object representing a pull
+Returns: dictionary containing all data of a pull
    
 
    
 
 Expand source code
 
 
    def getPullFeatures(self, pull):
+    """
+    Method to get all data for a particular pull. Parameters: pull - object representing a pull
+    Returns: dictionary containing all data of a pull
+    """
     pull_dict = {}
     pull_dict["Number"] = pull.number
     pull_dict["Title"] = pull.title
@@ -420,12 +453,17 @@ 
    Methods
    
 def getPullsByState(self, state)
     
 
    
-
    
+
    Extract pulls with given state. Parameters: state - state of the pull (open or closed)
+Return: list of dictionaries containing data regardining each pull
    
 
    
 
 Expand source code
 
 
    def getPullsByState(self, state):
+    """
+    Extract pulls with given state. Parameters: state - state of the pull (open or closed)
+    Return: list of dictionaries containing data regardining each pull
+    """
     pull_data = []
     try:  # Call the Github api to get all pulls
         pulls = self.current_repo.get_pulls(state=state, sort='create')
@@ -448,12 +486,17 @@ 
    Methods
    
 def listOfComments(self, comments)
     
 
    
-
    
+
    Method to form a list of json strings rerpesenting comments (reviews or issue).
+Parameters: comments - list of comment objects. Returns: List of json strings
    
 
    
 
 Expand source code
 
 
    def listOfComments(self, comments):
+    """
+    Method to form a list of json strings rerpesenting comments (reviews or issue).
+    Parameters: comments - list of comment objects. Returns: List of json strings
+    """
     list_comments = []
 
     # Iterate over each comment
@@ -500,12 +543,18 @@ 
    Methods
    
 def openRepo(self, organization, repo_name)
     
 
    
-
    
+
    Method to
+open (access) repository with given organization and repository name (reponame).
+Parameters: username - owner of the repository, repo_name - name of repo to be opened
    
 
    
 
 Expand source code
 
 
    def openRepo(self, organization, repo_name):
+    """
+    Method to  open (access) repository with given organization and repository name (reponame).
+    Parameters: username - owner of the repository, repo_name - name of repo to be opened
+    """
     self.current_repo = self.g_ses.get_repo(organization + "/" + repo_name)  # Open repo
     self.repo_opened = True
     self.repo_name = repo_name
diff --git a/docs/ml-conversational-analytic-tool/preProcessedDataset.html b/docs/ml-conversational-analytic-tool/preProcessedDataset.html
index 4f2d49c..c810642 100644
--- a/docs/ml-conversational-analytic-tool/preProcessedDataset.html
+++ b/docs/ml-conversational-analytic-tool/preProcessedDataset.html
@@ -36,10 +36,11 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
 import tensorflow as tf
 
 
-# Class to prepare dataset for machine learning tasks
 class PreProcessedDataset:
-    # Set flags and instance variables in constructor
     def __init__(self, vocab_size=1000, no_tokens=512, max_pull_length=100):
+        """
+        Set flags and instance variables in constructor
+        """
         self.annotated_data = None
         self.dataset = None
         self.full_dataset = None
@@ -59,8 +60,11 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
         self.full_dataset_ready = False
         self.encode_ready = False
 
-    # Setup the preprocessed dataset
     def setupPreProcess(self, annotated_filename, dataset_filename):
+        """
+        Setup the preprocessed dataset
+        """
+
         # Load datasets
         self.loadAnnotatedData(annotated_filename)
         self.loadDataset(dataset_filename)
@@ -79,8 +83,11 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
         self.full_dataset['Review_Comments'] = self.full_dataset['Review_Comments'].apply(
             lambda comment: stringToDict(comment))
 
-    # Encode all utterances
     def encodeData(self):
+        """
+        Encode all utterances
+        """
+
         all_utterances = self._getObsResUsers()
         # Encode each utterance
         for utterances in all_utterances:
@@ -98,8 +105,11 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
             self.results['Inclusive'] = self.results['Inclusive'][:idx] + self.results['Inclusive'][idx + 1:]
         self.curr_max_length = max([len(x) for x in self.all_encoded_utterances])
 
-    # Get matrix observation and results for ML task soutcome = Inclusive, Constructive, or None -> both 
     def getRoleAgnosticMatrix(self, outcome=None, padPull=True):
+        """
+        Get matrix observation and results for ML task outcome = Inclusive, Constructive, or None -> both
+        """
+
         obs = []
         for i in range(len(self.all_encoded_utterances)):
             def pad(inp):
@@ -116,17 +126,21 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
         res = self.getRes(outcome)
         return obs, res
 
-    # Get stacked matrix observation and results for ML task 
     def getRoleMatrix(self, outcome=None, padPull=True):
+        """
+        Get stacked matrix observation and results for ML task
+        """
         # Check if results must be padded to same length for each pull
         if padPull:
             # Author and reviwer layers
             layer_writer = []
             layer_reviewer = []
+
             for i in range(len(self.all_encoded_utterances)):
                 writer_comments = []
                 reviewer_comments = []
                 writer = self.all_users[i][0]
+
                 for j in range(len(self.all_encoded_utterances[i])):
                     # Check if author or reviewer
                     if self.all_users[i][j] == writer:
@@ -136,12 +150,14 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
                         reviewer_comments.append(self.all_encoded_utterances[i][j])
                         writer_comments.append(np.zeros(self.no_tokens))
                     padding = [0] * self.no_tokens
+
                     # Pad both reviewer and writer layers
                     while len(writer_comments) < self.curr_max_length:
                         writer_comments.append(padding.copy())
                         reviewer_comments.append(padding.copy())
                     layer_writer.append(np.array(writer_comments))
                     layer_reviewer.append(np.array(reviewer_comments))
+
             # Stack reviwer and author matrices
             obs = np.stack((layer_writer, layer_reviewer), axis=3)
             res = self.getRes(outcome)
@@ -152,6 +168,7 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
                 writer_comments = []
                 reviewer_comments = []
                 writer = self.all_users[i][0]
+
                 for j in range(len(self.all_encoded_utterances[i])):
                     if self.all_users[i][j] == writer:
                         writer_comments.append(self.all_encoded_utterances[i][j])
@@ -164,24 +181,32 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
             res = self.getRes(outcome)
             return obs, res
 
-    # Get list of results
     def getRes(self, outcome=None):
+        """
+        Get list of results
+        """
         if outcome:
             return self.results[outcome]
         return pd.DataFrame(data=self.results)
 
-    # Load annotated dataset from file
     def loadAnnotatedData(self, filename):
+        """
+        Load annotated dataset from file
+        """
         self.annotated_data = pd.read_csv(filename)
         self.annotated_data_open = True
 
-    # Load raw data from file
     def loadDataset(self, filename):
+        """
+        Load raw data from file
+        """
         self.dataset = pd.read_csv(filename)
         self.dataset_open = True
 
-    # Encode an utterance through the lookup dictionary
     def encode(self, utterances):
+        """
+        Encode an utterance through the lookup dictionary
+        """
         encoded_utterances = []
         for utterance in utterances:
             obs = []
@@ -200,8 +225,10 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
             encoded_utterances.append(np.array(obs))
         return np.array(encoded_utterances)
 
-    # Setup the lookup dictionary through frequency encoding
     def _setupEncode(self):
+        """
+        Setup the lookup dictionary through frequency encoding
+        """
         word_counts = {}
         for index, row in self.annotated_data.iterrows():
             words = row['Thread'].split(" ")
@@ -210,6 +237,7 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
                     word_counts[word] = word_counts[word] + 1
                 else:
                     word_counts[word] = 1
+
         # Sort the words to generate code
         # Lower number : higher count
         # 0 - Padding, 1 - End, 2 - Missing
@@ -224,8 +252,10 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
                 break
         self.encode_ready = True
 
-    # Get all utterances by parsing dictionary for each pull
     def _getObsResUsers(self):
+        """
+        Get all utterances by parsing dictionary for each pull
+        """
         all_utterances = []
         for index, row in self.full_dataset.iterrows():
             utterances = []
@@ -234,17 +264,21 @@ 
    Module ml-conversational-analytic-tool.preProcessedDatas
             users.append(row['User'])
             temp_df_comments = pd.DataFrame(row['Comments'])
             temp_df_review_comments = pd.DataFrame(row["Review_Comments"])
+
             if len(temp_df_comments) > 0 or len(temp_df_review_comments) > 0:
                 all_comments = temp_df_comments.append(temp_df_review_comments)
                 all_comments['Created_At'] = pd.to_datetime(all_comments['Created_At'])
                 all_comments = all_comments.sort_values(by=['Created_At'])
+
                 for comment_index, comment_row in all_comments.iterrows():
                     utterances.append(comment_row['Body'])
                     users.append(comment_row['User'])
+
             all_utterances.append(utterances)
             self.all_users.append(users)
             self.results['Constructive'].append(row['Constructive'])
             self.results['Inclusive'].append(row['Inclusive'])
+
         return all_utterances
    
 
    
 
@@ -262,14 +296,16 @@ 
    Classes
    
 (vocab_size=1000, no_tokens=512, max_pull_length=100)
     
 
    
-
    
+
    Set flags and instance variables in constructor
    
 
    
 
 Expand source code
 
 
    class PreProcessedDataset:
-    # Set flags and instance variables in constructor
     def __init__(self, vocab_size=1000, no_tokens=512, max_pull_length=100):
+        """
+        Set flags and instance variables in constructor
+        """
         self.annotated_data = None
         self.dataset = None
         self.full_dataset = None
@@ -289,8 +325,11 @@ 
    Classes
    
         self.full_dataset_ready = False
         self.encode_ready = False
 
-    # Setup the preprocessed dataset
     def setupPreProcess(self, annotated_filename, dataset_filename):
+        """
+        Setup the preprocessed dataset
+        """
+
         # Load datasets
         self.loadAnnotatedData(annotated_filename)
         self.loadDataset(dataset_filename)
@@ -309,8 +348,11 @@ 
    Classes
    
         self.full_dataset['Review_Comments'] = self.full_dataset['Review_Comments'].apply(
             lambda comment: stringToDict(comment))
 
-    # Encode all utterances
     def encodeData(self):
+        """
+        Encode all utterances
+        """
+
         all_utterances = self._getObsResUsers()
         # Encode each utterance
         for utterances in all_utterances:
@@ -328,8 +370,11 @@ 
    Classes
    
             self.results['Inclusive'] = self.results['Inclusive'][:idx] + self.results['Inclusive'][idx + 1:]
         self.curr_max_length = max([len(x) for x in self.all_encoded_utterances])
 
-    # Get matrix observation and results for ML task soutcome = Inclusive, Constructive, or None -> both 
     def getRoleAgnosticMatrix(self, outcome=None, padPull=True):
+        """
+        Get matrix observation and results for ML task outcome = Inclusive, Constructive, or None -> both
+        """
+
         obs = []
         for i in range(len(self.all_encoded_utterances)):
             def pad(inp):
@@ -346,17 +391,21 @@ 
    Classes
    
         res = self.getRes(outcome)
         return obs, res
 
-    # Get stacked matrix observation and results for ML task 
     def getRoleMatrix(self, outcome=None, padPull=True):
+        """
+        Get stacked matrix observation and results for ML task
+        """
         # Check if results must be padded to same length for each pull
         if padPull:
             # Author and reviwer layers
             layer_writer = []
             layer_reviewer = []
+
             for i in range(len(self.all_encoded_utterances)):
                 writer_comments = []
                 reviewer_comments = []
                 writer = self.all_users[i][0]
+
                 for j in range(len(self.all_encoded_utterances[i])):
                     # Check if author or reviewer
                     if self.all_users[i][j] == writer:
@@ -366,12 +415,14 @@ 
    Classes
    
                         reviewer_comments.append(self.all_encoded_utterances[i][j])
                         writer_comments.append(np.zeros(self.no_tokens))
                     padding = [0] * self.no_tokens
+
                     # Pad both reviewer and writer layers
                     while len(writer_comments) < self.curr_max_length:
                         writer_comments.append(padding.copy())
                         reviewer_comments.append(padding.copy())
                     layer_writer.append(np.array(writer_comments))
                     layer_reviewer.append(np.array(reviewer_comments))
+
             # Stack reviwer and author matrices
             obs = np.stack((layer_writer, layer_reviewer), axis=3)
             res = self.getRes(outcome)
@@ -382,6 +433,7 @@ 
    Classes
    
                 writer_comments = []
                 reviewer_comments = []
                 writer = self.all_users[i][0]
+
                 for j in range(len(self.all_encoded_utterances[i])):
                     if self.all_users[i][j] == writer:
                         writer_comments.append(self.all_encoded_utterances[i][j])
@@ -394,24 +446,32 @@ 
    Classes
    
             res = self.getRes(outcome)
             return obs, res
 
-    # Get list of results
     def getRes(self, outcome=None):
+        """
+        Get list of results
+        """
         if outcome:
             return self.results[outcome]
         return pd.DataFrame(data=self.results)
 
-    # Load annotated dataset from file
     def loadAnnotatedData(self, filename):
+        """
+        Load annotated dataset from file
+        """
         self.annotated_data = pd.read_csv(filename)
         self.annotated_data_open = True
 
-    # Load raw data from file
     def loadDataset(self, filename):
+        """
+        Load raw data from file
+        """
         self.dataset = pd.read_csv(filename)
         self.dataset_open = True
 
-    # Encode an utterance through the lookup dictionary
     def encode(self, utterances):
+        """
+        Encode an utterance through the lookup dictionary
+        """
         encoded_utterances = []
         for utterance in utterances:
             obs = []
@@ -430,8 +490,10 @@ 
    Classes
    
             encoded_utterances.append(np.array(obs))
         return np.array(encoded_utterances)
 
-    # Setup the lookup dictionary through frequency encoding
     def _setupEncode(self):
+        """
+        Setup the lookup dictionary through frequency encoding
+        """
         word_counts = {}
         for index, row in self.annotated_data.iterrows():
             words = row['Thread'].split(" ")
@@ -440,6 +502,7 @@ 
    Classes
    
                     word_counts[word] = word_counts[word] + 1
                 else:
                     word_counts[word] = 1
+
         # Sort the words to generate code
         # Lower number : higher count
         # 0 - Padding, 1 - End, 2 - Missing
@@ -454,8 +517,10 @@ 
    Classes
    
                 break
         self.encode_ready = True
 
-    # Get all utterances by parsing dictionary for each pull
     def _getObsResUsers(self):
+        """
+        Get all utterances by parsing dictionary for each pull
+        """
         all_utterances = []
         for index, row in self.full_dataset.iterrows():
             utterances = []
@@ -464,17 +529,21 @@ 
    Classes
    
             users.append(row['User'])
             temp_df_comments = pd.DataFrame(row['Comments'])
             temp_df_review_comments = pd.DataFrame(row["Review_Comments"])
+
             if len(temp_df_comments) > 0 or len(temp_df_review_comments) > 0:
                 all_comments = temp_df_comments.append(temp_df_review_comments)
                 all_comments['Created_At'] = pd.to_datetime(all_comments['Created_At'])
                 all_comments = all_comments.sort_values(by=['Created_At'])
+
                 for comment_index, comment_row in all_comments.iterrows():
                     utterances.append(comment_row['Body'])
                     users.append(comment_row['User'])
+
             all_utterances.append(utterances)
             self.all_users.append(users)
             self.results['Constructive'].append(row['Constructive'])
             self.results['Inclusive'].append(row['Inclusive'])
+
         return all_utterances
    
 
    
 
    Methods
    
@@ -483,12 +552,15 @@ 
    Methods
    
 def encode(self, utterances)
     
 
    
-
    
+
    Encode an utterance through the lookup dictionary
    
 
    
 
 Expand source code
 
 
    def encode(self, utterances):
+    """
+    Encode an utterance through the lookup dictionary
+    """
     encoded_utterances = []
     for utterance in utterances:
         obs = []
@@ -512,12 +584,16 @@ 
    Methods
    
 def encodeData(self)
     
 
    
-
    
+
    Encode all utterances
    
 
    
 
 Expand source code
 
 
    def encodeData(self):
+    """
+    Encode all utterances
+    """
+
     all_utterances = self._getObsResUsers()
     # Encode each utterance
     for utterances in all_utterances:
@@ -540,12 +616,15 @@ 
    Methods
    
 def getRes(self, outcome=None)
     
 
    
-
    
+
    Get list of results
    
 
    
 
 Expand source code
 
 
    def getRes(self, outcome=None):
+    """
+    Get list of results
+    """
     if outcome:
         return self.results[outcome]
     return pd.DataFrame(data=self.results)
    
@@ -555,12 +634,16 @@ 
    Methods
    
 def getRoleAgnosticMatrix(self, outcome=None, padPull=True)
     
 
    
-
    
+
    Get matrix observation and results for ML task outcome = Inclusive, Constructive, or None -> both
    
 
    
 
 Expand source code
 
 
    def getRoleAgnosticMatrix(self, outcome=None, padPull=True):
+    """
+    Get matrix observation and results for ML task outcome = Inclusive, Constructive, or None -> both
+    """
+
     obs = []
     for i in range(len(self.all_encoded_utterances)):
         def pad(inp):
@@ -582,21 +665,26 @@ 
    Methods
    
 def getRoleMatrix(self, outcome=None, padPull=True)
     
 
    
-
    
+
    Get stacked matrix observation and results for ML task
    
 
    
 
 Expand source code
 
 
    def getRoleMatrix(self, outcome=None, padPull=True):
+    """
+    Get stacked matrix observation and results for ML task
+    """
     # Check if results must be padded to same length for each pull
     if padPull:
         # Author and reviwer layers
         layer_writer = []
         layer_reviewer = []
+
         for i in range(len(self.all_encoded_utterances)):
             writer_comments = []
             reviewer_comments = []
             writer = self.all_users[i][0]
+
             for j in range(len(self.all_encoded_utterances[i])):
                 # Check if author or reviewer
                 if self.all_users[i][j] == writer:
@@ -606,12 +694,14 @@ 
    Methods
    
                     reviewer_comments.append(self.all_encoded_utterances[i][j])
                     writer_comments.append(np.zeros(self.no_tokens))
                 padding = [0] * self.no_tokens
+
                 # Pad both reviewer and writer layers
                 while len(writer_comments) < self.curr_max_length:
                     writer_comments.append(padding.copy())
                     reviewer_comments.append(padding.copy())
                 layer_writer.append(np.array(writer_comments))
                 layer_reviewer.append(np.array(reviewer_comments))
+
         # Stack reviwer and author matrices
         obs = np.stack((layer_writer, layer_reviewer), axis=3)
         res = self.getRes(outcome)
@@ -622,6 +712,7 @@ 
    Methods
    
             writer_comments = []
             reviewer_comments = []
             writer = self.all_users[i][0]
+
             for j in range(len(self.all_encoded_utterances[i])):
                 if self.all_users[i][j] == writer:
                     writer_comments.append(self.all_encoded_utterances[i][j])
@@ -639,12 +730,15 @@ 
    Methods
    
 def loadAnnotatedData(self, filename)
     
 
    
-
    
+
    Load annotated dataset from file
    
 
    
 
 Expand source code
 
 
    def loadAnnotatedData(self, filename):
+    """
+    Load annotated dataset from file
+    """
     self.annotated_data = pd.read_csv(filename)
     self.annotated_data_open = True
    
 
    
@@ -653,12 +747,15 @@ 
    Methods
    
 def loadDataset(self, filename)
 
 
    
-
    
+
    Load raw data from file
    
 
    
 
 Expand source code
 
 
    def loadDataset(self, filename):
+    """
+    Load raw data from file
+    """
     self.dataset = pd.read_csv(filename)
     self.dataset_open = True
    
 
    
@@ -667,12 +764,16 @@ 
    Methods
    
 def setupPreProcess(self, annotated_filename, dataset_filename)
 
 
    
-
    
+
    Setup the preprocessed dataset
    
 
    
 
 Expand source code
 
 
    def setupPreProcess(self, annotated_filename, dataset_filename):
+    """
+    Setup the preprocessed dataset
+    """
+
     # Load datasets
     self.loadAnnotatedData(annotated_filename)
     self.loadDataset(dataset_filename)
diff --git a/docs/ml-conversational-analytic-tool/run.html b/docs/ml-conversational-analytic-tool/run.html
index 396398b..1c8639a 100644
--- a/docs/ml-conversational-analytic-tool/run.html
+++ b/docs/ml-conversational-analytic-tool/run.html
@@ -30,6 +30,8 @@ 
    Module ml-conversational-analytic-tool.run
    
 # SPDX-License-Identifier: Apache-2.0
 
 import argparse
+import tarfile
+import os
 
 from sklearn.model_selection import train_test_split
 
@@ -37,27 +39,39 @@ 
    Module ml-conversational-analytic-tool.run
    
 from baseLSTM import BaseLSTM
 from preProcessedDataset import PreProcessedDataset
 
-
-def run(annotated_filename, dataset_filename, outcome, encoding_type, model_type, padding):
+model_directory = 'models'
+
+def save_model(model, name, version):
+    if not os.path.exists(model_directory):
+        os.makedirs(model_directory)
+    model_path = "{}/{}".format(model_directory, name)
+    tar_file_name = "{}-{}.tar.gz".format(name, version)
+    model.saveModel(name=model_path, version=version)
+    os.chdir(model_path)
+    tar = tarfile.open(tar_file_name, "w:gz")
+    tar.add(version)
+    tar.close()
+    os.chdir("../../")
+    print("Model saved in {}/{}; {}/{}".format(model_path, version, model_path, tar_file_name))
+
+def run(annotated_filename, dataset_filename, outcome, encoding_type, model_type, padding, save_name, model_ver):
     # Setup dataset
     data = PreProcessedDataset()
     data.setupPreProcess(annotated_filename, dataset_filename)
     data.encodeData()
 
-    # Get data for training
-    if encoding_type == 'role':
-        obs, res = data.getRoleMatrix(outcome, padding)
-    elif encoding_type == 'role-agnostic':
-        obs, res = data.getRoleAgnosticMatrix(outcome, padding)
-
     # Create models
-    if model_type == 'CNN':
-        model = BaseCNN()
-    elif model_type == 'LSTM':
+    if model_type == 'LSTM':
         model = BaseLSTM()
+    else:
+        model = BaseCNN()
+
+    # Get data for training
     if encoding_type == 'role':
+        obs, res = data.getRoleMatrix(outcome, padding)
         model.makeModel2D(obs[0].shape)
-    elif encoding_type == 'role-agnostic':
+    else:
+        obs, res = data.getRoleAgnosticMatrix(outcome, padding)
         model.makeModel(obs[0].shape)
 
     # Train model
@@ -66,6 +80,10 @@ 
    Module ml-conversational-analytic-tool.run
    
 
     # Score model
     scores = model.scoreModel(test_obs, test_res)
+    
+    # Save model
+    if save_name is not None and len(save_name) > 0:
+        save_model(model=model, name=save_name+"-"+outcome, version=model_ver)
 
     return scores
 
@@ -75,27 +93,36 @@ 
    Module ml-conversational-analytic-tool.run
    
         description="Obtain models to determine constructive and inclusive feedback in Open source communities")
     parser.add_argument('annotated_filename', help='File location of annotated file')
     parser.add_argument('dataset_filename', help='File location of extracted dataset')
-    parser.add_argument('model', help='Model type to use for training')
+    parser.add_argument('model', help='Model type to use for training, supported CNN and LSTM')
     parser.add_argument('outcome', help='Inclusive, Constructive, or Both')
+    parser.add_argument('-save', metavar='NAME', help='Save the model using given NAME')
+    parser.add_argument('-save_version', metavar='VERSION', default='001',
+                        help='Together with -save NAME: save the model using given NAME and VERSION. '\
+                             'If omitted, 001 is used. The parameter is ignored if -save is missing.')
     parser.add_argument('-roleRelevant', action='store_true', default=False,
                         help='Encoding method differentiates b/w conversation roles')
     parser.add_argument('-pad', action='store_true', default=False, help='Pad total length of each pull')
 
     args = parser.parse_args()
+
+    if args.model != 'CNN' and args.model != 'LSTM':
+        raise Exception("Model must be either CNN or LSTM")
+
     encodingType = 'role'
     if not args.roleRelevant:
         encodingType = 'role-agnostic'
 
     if args.outcome != 'Both':
-        run_res = run(args.annotated_filename, args.dataset_filename, args.outcome, encodingType, args.model, args.pad)
+        run_res = run(args.annotated_filename, args.dataset_filename, args.outcome, encodingType,
+                      args.model, args.pad, args.save, args.save_version)
         print(run_res)
     else:
         run_res_constructive = run(args.annotated_filename, args.dataset_filename, 'Constructive', encodingType,
-                                   args.model, args.pad)
+                                   args.model, args.pad, args.save, args.save_version)
         print("Constructive: {}".format(run_res_constructive))
 
-        run_res_inclusive = run(args.annotated_filename, args.dataset_filename, 'Inclusive', encodingType, args.model,
-                                args.pad)
+        run_res_inclusive = run(args.annotated_filename, args.dataset_filename, 'Inclusive', encodingType,
+                                args.model, args.pad, args.save, args.save_version)
         print("Inclusvie: {}".format(run_res_inclusive))
    
 
    
 
@@ -107,7 +134,7 @@ 
    Module ml-conversational-analytic-tool.run
    
 
    Functions
    
 
    
 
    
-def run(annotated_filename, dataset_filename, outcome, encoding_type, model_type, padding)
+def run(annotated_filename, dataset_filename, outcome, encoding_type, model_type, padding, save_name, model_ver)
 
    
 
    
 
    
@@ -115,26 +142,24 @@ 
    Functions
    
 
 Expand source code
 
-
    def run(annotated_filename, dataset_filename, outcome, encoding_type, model_type, padding):
+
    def run(annotated_filename, dataset_filename, outcome, encoding_type, model_type, padding, save_name, model_ver):
     # Setup dataset
     data = PreProcessedDataset()
     data.setupPreProcess(annotated_filename, dataset_filename)
     data.encodeData()
 
-    # Get data for training
-    if encoding_type == 'role':
-        obs, res = data.getRoleMatrix(outcome, padding)
-    elif encoding_type == 'role-agnostic':
-        obs, res = data.getRoleAgnosticMatrix(outcome, padding)
-
     # Create models
-    if model_type == 'CNN':
-        model = BaseCNN()
-    elif model_type == 'LSTM':
+    if model_type == 'LSTM':
         model = BaseLSTM()
+    else:
+        model = BaseCNN()
+
+    # Get data for training
     if encoding_type == 'role':
+        obs, res = data.getRoleMatrix(outcome, padding)
         model.makeModel2D(obs[0].shape)
-    elif encoding_type == 'role-agnostic':
+    else:
+        obs, res = data.getRoleAgnosticMatrix(outcome, padding)
         model.makeModel(obs[0].shape)
 
     # Train model
@@ -143,10 +168,37 @@ 
    Functions
    
 
     # Score model
     scores = model.scoreModel(test_obs, test_res)
+    
+    # Save model
+    if save_name is not None and len(save_name) > 0:
+        save_model(model=model, name=save_name+"-"+outcome, version=model_ver)
 
     return scores
    
 
    
 
    
+
    
+def save_model(model, name, version)
+
    
+
    
+
    
+
    
+
+Expand source code
+
+
    def save_model(model, name, version):
+    if not os.path.exists(model_directory):
+        os.makedirs(model_directory)
+    model_path = "{}/{}".format(model_directory, name)
+    tar_file_name = "{}-{}.tar.gz".format(name, version)
+    model.saveModel(name=model_path, version=version)
+    os.chdir(model_path)
+    tar = tarfile.open(tar_file_name, "w:gz")
+    tar.add(version)
+    tar.close()
+    os.chdir("../../")
+    print("Model saved in {}/{}; {}/{}".format(model_path, version, model_path, tar_file_name))
    
+
    
+
    
 
 
 
    
@@ -166,6 +218,7 @@ 
    Index
    
 
  • Functions
    
 
 
    • 
 
diff --git a/docs/ml-conversational-analytic-tool/runDataExtraction.html b/docs/ml-conversational-analytic-tool/runDataExtraction.html
index 1ebe4fa..8d7d03b 100644
--- a/docs/ml-conversational-analytic-tool/runDataExtraction.html
+++ b/docs/ml-conversational-analytic-tool/runDataExtraction.html
@@ -35,8 +35,10 @@ 
    Module ml-conversational-analytic-tool.runDataExtraction
 from githubDataExtraction import GithubDataExtractor
 
 
-# Method to extract data for all repositories in organization
 def getRepos(access_token, organization, reaction):
+    """
+    Method to extract data for all repositories in organization
+    """
     extractor = GithubDataExtractor(access_token)  # Create object
     repos = extractor.g_ses.get_organization(organization).get_repos()
     for repo in repos:
@@ -45,8 +47,10 @@ 
    Module ml-conversational-analytic-tool.runDataExtraction
         extractor.getAllPulls("", reaction)
 
 
-# Method to extract data for an individual repository
 def getRepo(access_token, organization, reponame, reaction):
+    """
+    Method to extract data for an individual repository
+    """
     extractor = GithubDataExtractor(access_token)  # Create object 
     extractor.openRepo(organization, reponame)
     extractor.getAllPulls("", reaction)
@@ -77,12 +81,15 @@ 
    Functions
    
 def getRepo(access_token, organization, reponame, reaction)
     
 
    
-
    
+
    Method to extract data for an individual repository
    
 
    
 
 Expand source code
 
 
    def getRepo(access_token, organization, reponame, reaction):
+    """
+    Method to extract data for an individual repository
+    """
     extractor = GithubDataExtractor(access_token)  # Create object 
     extractor.openRepo(organization, reponame)
     extractor.getAllPulls("", reaction)
    
@@ -92,12 +99,15 @@ 
    Functions
    
 def getRepos(access_token, organization, reaction)
     
 
    
-
    
+
    Method to extract data for all repositories in organization
    
 
    
 
 Expand source code
 
 
    def getRepos(access_token, organization, reaction):
+    """
+    Method to extract data for all repositories in organization
+    """
     extractor = GithubDataExtractor(access_token)  # Create object
     repos = extractor.g_ses.get_organization(organization).get_repos()
     for repo in repos: