predicting-credit-risk-model-pipeline.ipynb

{"cells":[{"metadata":{"_uuid":"7f1516708a5b791610a2eeae594c5b2f25d55ac3","_cell_guid":"96611c58-d3d2-4fdb-849a-54b0cd5c72e1"},"cell_type":"markdown","source":"<h3>- Note that I will use the Dataset with the Risk column </h3>"},{"metadata":{"_uuid":"c4bdcc85e7b480e52ff70dfd20092951cfb776ea","_cell_guid":"bcce07c4-1ef6-4db6-8170-496367f18ce8"},"cell_type":"markdown","source":"<h1>Welcome to my Kernel ! </h1>\n\n<h2>I will do some Explorations through the German Credit Risk to understand their distribuitions and patterns. </h2>"},{"metadata":{"_uuid":"f04d8fb1ecb68e4d0349de77e8d90fd50bfe473c","_cell_guid":"62f7266e-1c41-4020-a0e6-f5b46151b0df"},"cell_type":"markdown","source":"Look for another interesting Kernels on https://www.kaggle.com/kabure/kernels<br>\nGive me your feedback and if yo like this kernel, <b>votes up</b>"},{"metadata":{"_uuid":"104a4548bc92aff6a309f89ef63b0214f350da22","_cell_guid":"ae98792a-c30c-45c6-b9c4-6ee28df4a41a"},"cell_type":"markdown","source":"# Tables of Content:\n\n**1. [Introduction](#Introduction)** <br>\n    - Info's about datasets\n**2. [Librarys](#Librarys)** <br>\n    - Importing Librarys\n    - Importing Dataset\n**3. [Knowning the data](#Known)** <br>\n    - 3.1 Looking the Type of Data\n    - 3.2 Shape of data\n    - 3.3 Null Numbers\n    - 3.4 Unique values\n    - 3.5 The first rows of our dataset\n**4. [Exploring some Variables](#Explorations)** <br>\n    - 4.1 Ploting some graphical and descriptive informations\n**5. [Correlation of data](#Correlation)** <br>\n\t- 5.1 Correlation Data\n**6. [Preprocess](#Preprocessing)** <br>\n\t- 6.1 Importing Librarys\n\t- 6.2 Setting X and Y\n    - 6.3 Spliting the X and Y in train and test \n**7. 1 [Model 1](#Modelling 1)** <br>\n    - 7.1.1 Random Forest \n    - 7.1.2 Score values\n    - 7.1.3 Cross Validation \n**7. 2 [Model 2](#Modelling 2)** <br>\n    - 7.2.1 Logistic Regression \n    - 7.2.2 Score values\n    - 7.2.3 Cross Validation \n    - 7.2.4 ROC Curve"},{"metadata":{"_uuid":"240bb7db644ecf54bc1a12bc1f0f41164415ab54","_cell_guid":"cc658486-a1b5-4b88-b340-bb4ffe268e5c"},"cell_type":"markdown","source":"<a id=\"Introduction\"></a> <br>\n"},{"metadata":{"_uuid":"54ffb8214adb51a4145611f2780082d3bab807e8","_cell_guid":"7d0ff485-f6eb-45d5-a138-2f1d640f4208"},"cell_type":"markdown","source":"# **1. Introduction:** \n<h2>Context</h2>\nThe original dataset contains 1000 entries with 20 categorial/symbolic attributes prepared by Prof. Hofmann. In this dataset, each entry represents a person who takes a credit by a bank. Each person is classified as good or bad credit risks according to the set of attributes. The link to the original dataset can be found below.\n\n<h2>Content</h2>\nIt is almost impossible to understand the original dataset due to its complicated system of categories and symbols. Thus, I wrote a small Python script to convert it into a readable CSV file. Several columns are simply ignored, because in my opinion either they are not important or their descriptions are obscure. The selected attributes are:\n\n<b>Age </b>(numeric)<br>\n<b>Sex </b>(text: male, female)<br>\n<b>Job </b>(numeric: 0 - unskilled and non-resident, 1 - unskilled and resident, 2 - skilled, 3 - highly skilled)<br>\n<b>Housing</b> (text: own, rent, or free)<br>\n<b>Saving accounts</b> (text - little, moderate, quite rich, rich)<br>\n<b>Checking account </b>(numeric, in DM - Deutsch Mark)<br>\n<b>Credit amount</b> (numeric, in DM)<br>\n<b>Duration</b> (numeric, in month)<br>\n<b>Purpose</b>(text: car, furniture/equipment, radio/TV, domestic appliances, repairs, education, business, vacation/others<br>\n<b>Risk </b> (Value target - Good or Bad Risk)<br>"},{"metadata":{"_uuid":"cdb97612a0a3756be77c572ee938b42401b54284","_cell_guid":"c136699c-701b-445f-8522-8ff3f1461b92"},"cell_type":"markdown","source":"<i>English is not my first language, so, sorry about any error</i>"},{"metadata":{"_uuid":"20fbb39e66aee438052d48c923b9f09bc8d651b1","_cell_guid":"b09532e1-9964-49e4-932e-7fe8ef29d9e5"},"cell_type":"markdown","source":"<a id=\"Librarys\"></a> <br>\n# **2. Librarys:** \n- Importing Librarys\n- Importing Dataset"},{"metadata":{"_uuid":"ebb00fd08333f52902981f381d94492ec2819bfd","_cell_guid":"98e8c140-b331-4a2e-98a1-aa76dd1ae9c3","trusted":true},"cell_type":"code","source":"#Load the librarys\nimport pandas as pd #To work with dataset\nimport numpy as np #Math library\nimport seaborn as sns #Graph library that use matplot in background\nimport matplotlib.pyplot as plt #to plot some parameters in seaborn\n\n#Importing the data\ndf_credit = pd.read_csv(\"../input/german-credit-data-with-risk/german_credit_data.csv\",index_col=0)","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"b1edd34feb9c638a57fe28bbb56695a9ece694c0","_cell_guid":"8f6af2bc-3f44-4134-bb3f-feaabcce6254"},"cell_type":"markdown","source":"<a id=\"Known\"></a> <br>\n# **3. First Look at the data:** \n- Looking the Type of Data\n- Null Numbers\n- Unique values\n- The first rows of our dataset"},{"metadata":{"_uuid":"017a2be77ff92841897d9b25e898927e641b90fb","_cell_guid":"81649261-de36-4f16-9816-643d90d3d2aa","trusted":true},"cell_type":"code","source":"#Searching for Missings,type of data and also known the shape of data\nprint(df_credit.info())","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"3e45297ce0530246c288cb8ab64378b639246f6b","_cell_guid":"379c0593-dbc5-4523-b130-8a7ee71bce5b","trusted":true},"cell_type":"code","source":"#Looking unique values\nprint(df_credit.nunique())\n#Looking the data\nprint(df_credit.head())","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"06a3ac5f2cc47753350a7d3dd8ad92873f03b16b","_cell_guid":"2c1da4dc-39d6-4f4a-9a3d-c7c31ddb5e57"},"cell_type":"markdown","source":"# **4. Some explorations:** <a id=\"Explorations\"></a> <br>\n\n- Starting by distribuition of column Age.\n- Some Seaborn graphical\n- Columns crossing\n\n"},{"metadata":{"_uuid":"71d1c34259be3ea63a76ac5010c96ccc90f17f40","_cell_guid":"46bd5d63-dfe1-477a-8647-926471e7ee37"},"cell_type":"markdown","source":"<h2>Let's start looking through target variable and their distribuition</h2>"},{"metadata":{"_uuid":"31cd32f3d652470e6a1e3f371778cfc6f557e34a","_kg_hide-input":true,"_cell_guid":"6c974adc-9d98-4648-ada5-15ff5f1b615c","trusted":true},"cell_type":"code","source":"# it's a library that we work with plotly\nimport plotly.offline as py \npy.init_notebook_mode(connected=True) # this code, allow us to work with offline plotly version\nimport plotly.graph_objs as go # it's like \"plt\" of matplot\nimport plotly.tools as tls # It's useful to we get some tools of plotly\nimport warnings # This library will be used to ignore some warnings\nfrom collections import Counter # To do counter of some features\n\ntrace0 = go.Bar(\n            x = df_credit[df_credit[\"Risk\"]== 'good'][\"Risk\"].value_counts().index.values,\n            y = df_credit[df_credit[\"Risk\"]== 'good'][\"Risk\"].value_counts().values,\n            name='Good credit'\n    )\n\ntrace1 = go.Bar(\n            x = df_credit[df_credit[\"Risk\"]== 'bad'][\"Risk\"].value_counts().index.values,\n            y = df_credit[df_credit[\"Risk\"]== 'bad'][\"Risk\"].value_counts().values,\n            name='Bad credit'\n    )\n\ndata = [trace0, trace1]\n\nlayout = go.Layout(\n    \n)\n\nlayout = go.Layout(\n    yaxis=dict(\n        title='Count'\n    ),\n    xaxis=dict(\n        title='Risk Variable'\n    ),\n    title='Target variable distribution'\n)\n\nfig = go.Figure(data=data, layout=layout)\n\npy.iplot(fig, filename='grouped-bar')","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"db83305148e0d597c7ec66899a770eeeeced6b33","_cell_guid":"f8e02b71-7445-4124-925c-15178141c2af"},"cell_type":"markdown","source":"I will try implement some interactive visuals in my Kernels, this will be the first, inspired in Alexader's Kernel and I will also continue implementing plotly and bokeh in my Kerne"},{"metadata":{"_uuid":"d45ccf0ea0715be8421e05e69ddcc22bd611e814","_kg_hide-input":true,"scrolled":false,"_cell_guid":"967e45bb-8518-49e0-afc3-2cb8ed424f4b","trusted":true},"cell_type":"code","source":"df_good = df_credit.loc[df_credit[\"Risk\"] == 'good']['Age'].values.tolist()\ndf_bad = df_credit.loc[df_credit[\"Risk\"] == 'bad']['Age'].values.tolist()\ndf_age = df_credit['Age'].values.tolist()\n\n#First plot\ntrace0 = go.Histogram(\n    x=df_good,\n    histnorm='probability',\n    name=\"Good Credit\"\n)\n#Second plot\ntrace1 = go.Histogram(\n    x=df_bad,\n    histnorm='probability',\n    name=\"Bad Credit\"\n)\n#Third plot\ntrace2 = go.Histogram(\n    x=df_age,\n    histnorm='probability',\n    name=\"Overall Age\"\n)\n\n#Creating the grid\nfig = tls.make_subplots(rows=2, cols=2, specs=[[{}, {}], [{'colspan': 2}, None]],\n                          subplot_titles=('Good','Bad', 'General Distribuition'))\n\n#setting the figs\nfig.append_trace(trace0, 1, 1)\nfig.append_trace(trace1, 1, 2)\nfig.append_trace(trace2, 2, 1)\n\nfig['layout'].update(showlegend=True, title='Age Distribuition', bargap=0.05)\npy.iplot(fig, filename='custom-sized-subplot-with-subplot-titles')","execution_count":null,"outputs":[]},{"metadata":{"_kg_hide-output":true,"_kg_hide-input":true,"scrolled":false,"_cell_guid":"5b7bff50-2d21-400d-919a-49acdc6bcd14","_uuid":"4a4939e2f985c12c80de727c9e3e24081f57ace8","trusted":true},"cell_type":"code","source":"df_good = df_credit[df_credit[\"Risk\"] == 'good']\ndf_bad = df_credit[df_credit[\"Risk\"] == 'bad']\n\nfig, ax = plt.subplots(nrows=2, figsize=(12,8))\nplt.subplots_adjust(hspace = 0.4, top = 0.8)\n\ng1 = sns.distplot(df_good[\"Age\"], ax=ax[0], \n             color=\"g\")\ng1 = sns.distplot(df_bad[\"Age\"], ax=ax[0], \n             color='r')\ng1.set_title(\"Age Distribuition\", fontsize=15)\ng1.set_xlabel(\"Age\")\ng1.set_xlabel(\"Frequency\")\n\ng2 = sns.countplot(x=\"Age\",data=df_credit, \n              palette=\"hls\", ax=ax[1], \n              hue = \"Risk\")\ng2.set_title(\"Age Counting by Risk\", fontsize=15)\ng2.set_xlabel(\"Age\")\ng2.set_xlabel(\"Count\")\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"4eaf4a2da2ea133fcb227313c72ab69f0f1328fa","_cell_guid":"dcfd7a3b-d43d-4c35-8b7c-cd0a3184881f"},"cell_type":"markdown","source":"<h2>Creating an categorical variable to handle with the Age variable </h2>"},{"metadata":{"_uuid":"e834b43cb46b510ca9d352297620e92b3f0f6db7","_kg_hide-output":false,"_kg_hide-input":false,"_cell_guid":"77674108-1042-4a9d-8820-decd3f4c04f6","trusted":true},"cell_type":"code","source":"#Let's look the Credit Amount column\ninterval = (18, 25, 35, 60, 120)\n\ncats = ['Student', 'Young', 'Adult', 'Senior']\ndf_credit[\"Age_cat\"] = pd.cut(df_credit.Age, interval, labels=cats)\n\n\ndf_good = df_credit[df_credit[\"Risk\"] == 'good']\ndf_bad = df_credit[df_credit[\"Risk\"] == 'bad']","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"f9fc969cd984d3f5a667456bf697db442b5215a9","_kg_hide-input":true,"scrolled":false,"_cell_guid":"052fc2b2-b2e2-4a13-81fd-3ac1b0c84a30","trusted":true},"cell_type":"code","source":"trace0 = go.Box(\n    y=df_good[\"Credit amount\"],\n    x=df_good[\"Age_cat\"],\n    name='Good credit',\n    marker=dict(\n        color='#3D9970'\n    )\n)\n\ntrace1 = go.Box(\n    y=df_bad['Credit amount'],\n    x=df_bad['Age_cat'],\n    name='Bad credit',\n    marker=dict(\n        color='#FF4136'\n    )\n)\n    \ndata = [trace0, trace1]\n\nlayout = go.Layout(\n    yaxis=dict(\n        title='Credit Amount (US Dollar)',\n        zeroline=False\n    ),\n    xaxis=dict(\n        title='Age Categorical'\n    ),\n    boxmode='group'\n)\nfig = go.Figure(data=data, layout=layout)\n\npy.iplot(fig, filename='box-age-cat')","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"16f8057c98761549f15c336976063b94e46843db","_cell_guid":"f30567d8-9826-4495-a735-e78f965b84cc"},"cell_type":"markdown","source":"Interesting distribuition"},{"metadata":{"_uuid":"a5e8088cd21325c077fe92b68246add9cc7aeb7f","_cell_guid":"39066b14-b573-4254-bc61-8b265e7bae5f"},"cell_type":"markdown","source":"<h2>I will now Look the distribuition of Housing own and rent by Risk</h2>\n"},{"metadata":{"_uuid":"e0b8b82f2090041fa13102caf82293e3e5bd015a","_kg_hide-input":true,"_cell_guid":"f04d566c-92f3-4745-84e8-63b87e6c4b4c","trusted":true},"cell_type":"code","source":"#First plot\ntrace0 = go.Bar(\n    x = df_credit[df_credit[\"Risk\"]== 'good'][\"Housing\"].value_counts().index.values,\n    y = df_credit[df_credit[\"Risk\"]== 'good'][\"Housing\"].value_counts().values,\n    name='Good credit'\n)\n\n#Second plot\ntrace1 = go.Bar(\n    x = df_credit[df_credit[\"Risk\"]== 'bad'][\"Housing\"].value_counts().index.values,\n    y = df_credit[df_credit[\"Risk\"]== 'bad'][\"Housing\"].value_counts().values,\n    name=\"Bad Credit\"\n)\n\ndata = [trace0, trace1]\n\nlayout = go.Layout(\n    title='Housing Distribuition'\n)\n\n\nfig = go.Figure(data=data, layout=layout)\n\npy.iplot(fig, filename='Housing-Grouped')","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"f0bbbfef78b2172aed6eec91a0082fb24a6b6652","_cell_guid":"85a1de27-ac07-4317-b2cb-f0ef0fc50263"},"cell_type":"markdown","source":"we can see that the own and good risk have a high correlation"},{"metadata":{"_uuid":"e14a1b38effd533bf963fce312ff5afde85d4e46","_cell_guid":"2ee9d053-57a7-4448-b00d-52f85b4b7435"},"cell_type":"markdown","source":"<h3>Distribuition of Credit Amount by Housing</h3>"},{"metadata":{"_uuid":"29da7e40ac47969c5fb2e747bfffcf1a78a795a0","_kg_hide-input":true,"scrolled":false,"_cell_guid":"6eac2d4d-2389-4001-816a-9346d76acaaf","trusted":true},"cell_type":"code","source":"fig = {\n    \"data\": [\n        {\n            \"type\": 'violin',\n            \"x\": df_good['Housing'],\n            \"y\": df_good['Credit amount'],\n            \"legendgroup\": 'Good Credit',\n            \"scalegroup\": 'No',\n            \"name\": 'Good Credit',\n            \"side\": 'negative',\n            \"box\": {\n                \"visible\": True\n            },\n            \"meanline\": {\n                \"visible\": True\n            },\n            \"line\": {\n                \"color\": 'blue'\n            }\n        },\n        {\n            \"type\": 'violin',\n            \"x\": df_bad['Housing'],\n            \"y\": df_bad['Credit amount'],\n            \"legendgroup\": 'Bad Credit',\n            \"scalegroup\": 'No',\n            \"name\": 'Bad Credit',\n            \"side\": 'positive',\n            \"box\": {\n                \"visible\": True\n            },\n            \"meanline\": {\n                \"visible\": True\n            },\n            \"line\": {\n                \"color\": 'green'\n            }\n        }\n    ],\n    \"layout\" : {\n        \"yaxis\": {\n            \"zeroline\": False,\n        },\n        \"violingap\": 0,\n        \"violinmode\": \"overlay\"\n    }\n}\n\n\npy.iplot(fig, filename = 'violin/split', validate = False)","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"1cfb55b3552be9e7dd67de9ed24c72af973cbe4b","_cell_guid":"c4b74894-53b4-4a93-83e1-761d3c7feb3a"},"cell_type":"markdown","source":"Interesting moviments! Highest values come from category \"free\" and we have a different distribuition by Risk"},{"metadata":{"_uuid":"a4b29db97c9da81196b5ef63dae3bc7dd5c18c48","_cell_guid":"bb043f80-4948-476b-a9ed-4d5850a08c2e"},"cell_type":"markdown","source":"<h2>Looking the diference by Sex</h2>"},{"metadata":{"_uuid":"7baf9501935c7ff1eee9cf519cea74ab3e858f98","_kg_hide-input":true,"_cell_guid":"be4c6440-bdf4-4051-acef-f6db9c43c7de","trusted":true},"cell_type":"code","source":"#First plot\ntrace0 = go.Bar(\n    x = df_credit[df_credit[\"Risk\"]== 'good'][\"Sex\"].value_counts().index.values,\n    y = df_credit[df_credit[\"Risk\"]== 'good'][\"Sex\"].value_counts().values,\n    name='Good credit'\n)\n\n#First plot 2\ntrace1 = go.Bar(\n    x = df_credit[df_credit[\"Risk\"]== 'bad'][\"Sex\"].value_counts().index.values,\n    y = df_credit[df_credit[\"Risk\"]== 'bad'][\"Sex\"].value_counts().values,\n    name=\"Bad Credit\"\n)\n\n#Second plot\ntrace2 = go.Box(\n    x = df_credit[df_credit[\"Risk\"]== 'good'][\"Sex\"],\n    y = df_credit[df_credit[\"Risk\"]== 'good'][\"Credit amount\"],\n    name=trace0.name\n)\n\n#Second plot 2\ntrace3 = go.Box(\n    x = df_credit[df_credit[\"Risk\"]== 'bad'][\"Sex\"],\n    y = df_credit[df_credit[\"Risk\"]== 'bad'][\"Credit amount\"],\n    name=trace1.name\n)\n\ndata = [trace0, trace1, trace2,trace3]\n\n\nfig = tls.make_subplots(rows=1, cols=2, \n                        subplot_titles=('Sex Count', 'Credit Amount by Sex'))\n\nfig.append_trace(trace0, 1, 1)\nfig.append_trace(trace1, 1, 1)\nfig.append_trace(trace2, 1, 2)\nfig.append_trace(trace3, 1, 2)\n\nfig['layout'].update(height=400, width=800, title='Sex Distribuition', boxmode='group')\npy.iplot(fig, filename='sex-subplot')","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"8bd5867db79d8b3fa561e84c02ab7694bdb23b04","_cell_guid":"6bb67a05-9d3a-4098-adce-44a5fb8daed1"},"cell_type":"markdown","source":"<b> How can I set the boxplots in different places? how can I use the same legend to both graphs?</b>"},{"metadata":{"_uuid":"f5a6c2dea040a31289910da2ae36c296559d5997","_cell_guid":"b8cf27c8-2170-401f-839d-a732108933e4"},"cell_type":"markdown","source":"I will create categories of Age and look the distribuition of Credit Amount by Risk...\n"},{"metadata":{"_uuid":"8345789033140b6b9f9f5a40ba8fd5426ee823e4","_cell_guid":"76dfaeb3-39ac-4777-8966-b854d2a66fdb"},"cell_type":"markdown","source":"I will do some explorations through the Job\n- Distribuition\n- Crossed by Credit amount\n- Crossed by Age"},{"metadata":{"_uuid":"bd5a104e256f8b1f567c12537a39662de6288db0","_kg_hide-input":true,"_cell_guid":"78075d47-5299-419c-b061-3bc4b17597e8","trusted":true},"cell_type":"code","source":"#First plot\ntrace0 = go.Bar(\n    x = df_credit[df_credit[\"Risk\"]== 'good'][\"Job\"].value_counts().index.values,\n    y = df_credit[df_credit[\"Risk\"]== 'good'][\"Job\"].value_counts().values,\n    name='Good credit Distribuition'\n)\n\n#Second plot\ntrace1 = go.Bar(\n    x = df_credit[df_credit[\"Risk\"]== 'bad'][\"Job\"].value_counts().index.values,\n    y = df_credit[df_credit[\"Risk\"]== 'bad'][\"Job\"].value_counts().values,\n    name=\"Bad Credit Distribuition\"\n)\n\ndata = [trace0, trace1]\n\nlayout = go.Layout(\n    title='Job Distribuition'\n)\n\nfig = go.Figure(data=data, layout=layout)\n\npy.iplot(fig, filename='grouped-bar')","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"db417f745474118fc75e5ba18637efb84877afaa","_kg_hide-input":true,"_cell_guid":"08f291c1-5e82-4fa6-9a25-bc25dec0f147","trusted":true},"cell_type":"code","source":"trace0 = go.Box(\n    x=df_good[\"Job\"],\n    y=df_good[\"Credit amount\"],\n    name='Good credit'\n)\n\ntrace1 = go.Box(\n    x=df_bad['Job'],\n    y=df_bad['Credit amount'],\n    name='Bad credit'\n)\n    \ndata = [trace0, trace1]\n\nlayout = go.Layout(\n    yaxis=dict(\n        title='Credit Amount distribuition by Job'\n    ),\n    boxmode='group'\n)\nfig = go.Figure(data=data, layout=layout)\n\npy.iplot(fig, filename='box-age-cat')","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"f807c70db19443844e048eea9d002736a074ea21","_kg_hide-input":true,"_cell_guid":"335dc135-ea52-4547-a7cd-e41391501ca5","trusted":true},"cell_type":"code","source":"\nfig = {\n    \"data\": [\n        {\n            \"type\": 'violin',\n            \"x\": df_good['Job'],\n            \"y\": df_good['Age'],\n            \"legendgroup\": 'Good Credit',\n            \"scalegroup\": 'No',\n            \"name\": 'Good Credit',\n            \"side\": 'negative',\n            \"box\": {\n                \"visible\": True\n            },\n            \"meanline\": {\n                \"visible\": True\n            },\n            \"line\": {\n                \"color\": 'blue'\n            }\n        },\n        {\n            \"type\": 'violin',\n            \"x\": df_bad['Job'],\n            \"y\": df_bad['Age'],\n            \"legendgroup\": 'Bad Credit',\n            \"scalegroup\": 'No',\n            \"name\": 'Bad Credit',\n            \"side\": 'positive',\n            \"box\": {\n                \"visible\": True\n            },\n            \"meanline\": {\n                \"visible\": True\n            },\n            \"line\": {\n                \"color\": 'green'\n            }\n        }\n    ],\n    \"layout\" : {\n        \"yaxis\": {\n            \"zeroline\": False,\n        },\n        \"violingap\": 0,\n        \"violinmode\": \"overlay\"\n    }\n}\n\n\npy.iplot(fig, filename = 'Age-Housing', validate = False)","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"ead3fef545ee735a5a17bb57bc432a9b0fec93f7","_kg_hide-output":true,"_kg_hide-input":true,"_cell_guid":"041709ef-5168-473d-ab55-d95644bfc07f","trusted":true},"cell_type":"code","source":"fig, ax = plt.subplots(figsize=(12,12), nrows=2)\n\ng1 = sns.boxplot(x=\"Job\", y=\"Credit amount\", data=df_credit, \n            palette=\"hls\", ax=ax[0], hue=\"Risk\")\ng1.set_title(\"Credit Amount by Job\", fontsize=15)\ng1.set_xlabel(\"Job Reference\", fontsize=12)\ng1.set_ylabel(\"Credit Amount\", fontsize=12)\n\ng2 = sns.violinplot(x=\"Job\", y=\"Age\", data=df_credit, ax=ax[1],  \n               hue=\"Risk\", split=True, palette=\"hls\")\ng2.set_title(\"Job Type reference x Age\", fontsize=15)\ng2.set_xlabel(\"Job Reference\", fontsize=12)\ng2.set_ylabel(\"Age\", fontsize=12)\n\nplt.subplots_adjust(hspace = 0.4,top = 0.9)\n\nplt.show()\n","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"c120123985f2f09dbab5d1be70e400ca55fe5b15","_cell_guid":"5ddc12f0-61c9-46fd-9ac5-72054b7e65b3"},"cell_type":"markdown","source":"Looking the distribuition of Credit Amont"},{"metadata":{"_kg_hide-output":false,"_kg_hide-input":true,"scrolled":false,"_cell_guid":"08186a5c-7f62-4f92-b2c0-7040143d63d1","_uuid":"e9820ae878fa287084608d982b26b0f0eb3925d2","trusted":true},"cell_type":"code","source":"import plotly.figure_factory as ff\n\nimport numpy as np\n\n# Add histogram data\nx1 = np.log(df_good['Credit amount']) \nx2 = np.log(df_bad[\"Credit amount\"])\n\n# Group data together\nhist_data = [x1, x2]\n\ngroup_labels = ['Good Credit', 'Bad Credit']\n\n# Create distplot with custom bin_size\nfig = ff.create_distplot(hist_data, group_labels, bin_size=.2)\n\n# Plot!\npy.iplot(fig, filename='Distplot with Multiple Datasets')","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"0e46f1f9d464c0223e45a7c089b0abd552965ef7","_kg_hide-output":true,"_kg_hide-input":true,"_cell_guid":"26b0bcb6-00f8-4116-85f5-7b26605b20d8","trusted":true},"cell_type":"code","source":"#Ploting the good and bad dataframes in distplot\nplt.figure(figsize = (8,5))\n\ng= sns.distplot(df_good['Credit amount'], color='r')\ng = sns.distplot(df_bad[\"Credit amount\"], color='g')\ng.set_title(\"Credit Amount Frequency distribuition\", fontsize=15)\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"f50b556b07c699f82314235d23a912fcc9a93371","_cell_guid":"7c43a4cd-cf0a-433c-9549-ee4edc46b8a9"},"cell_type":"markdown","source":"Distruibution of Saving accounts by Risk"},{"metadata":{"_uuid":"68625a64933accf95647b5a097d14508bc78a1e7","_kg_hide-input":true,"_cell_guid":"ba0ae362-7005-490b-b682-140fbed017b1","trusted":true},"cell_type":"code","source":"from plotly import tools\nimport numpy as np\nimport plotly.graph_objs as go\n\ncount_good = go.Bar(\n    x = df_good[\"Saving accounts\"].value_counts().index.values,\n    y = df_good[\"Saving accounts\"].value_counts().values,\n    name='Good credit'\n)\ncount_bad = go.Bar(\n    x = df_bad[\"Saving accounts\"].value_counts().index.values,\n    y = df_bad[\"Saving accounts\"].value_counts().values,\n    name='Bad credit'\n)\n\n\nbox_1 = go.Box(\n    x=df_good[\"Saving accounts\"],\n    y=df_good[\"Credit amount\"],\n    name='Good credit'\n)\nbox_2 = go.Box(\n    x=df_bad[\"Saving accounts\"],\n    y=df_bad[\"Credit amount\"],\n    name='Bad credit'\n)\n\nscat_1 = go.Box(\n    x=df_good[\"Saving accounts\"],\n    y=df_good[\"Age\"],\n    name='Good credit'\n)\nscat_2 = go.Box(\n    x=df_bad[\"Saving accounts\"],\n    y=df_bad[\"Age\"],\n    name='Bad credit'\n)\n\ndata = [scat_1, scat_2, box_1, box_2, count_good, count_bad]\n\nfig = tools.make_subplots(rows=2, cols=2, specs=[[{}, {}], [{'colspan': 2}, None]],\n                          subplot_titles=('Count Saving Accounts','Credit Amount by Savings Acc', \n                                          'Age by Saving accounts'))\n\nfig.append_trace(count_good, 1, 1)\nfig.append_trace(count_bad, 1, 1)\n\nfig.append_trace(box_2, 1, 2)\nfig.append_trace(box_1, 1, 2)\n\nfig.append_trace(scat_1, 2, 1)\nfig.append_trace(scat_2, 2, 1)\n\n\n\nfig['layout'].update(height=700, width=800, title='Saving Accounts Exploration', boxmode='group')\n\npy.iplot(fig, filename='combined-savings')\n","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"bc422a23f5310a21f16e4c44442afa75723d40c3","_cell_guid":"9a936151-e4f1-4297-a87e-7f2bf2e4d3d0"},"cell_type":"markdown","source":"How can I better configure the legends?  I am trying to substitute the graph below, so how can I use the violinplot on subplots of plotly?"},{"metadata":{"_uuid":"725a327525000b8951bba1ba36f0611c92e95c9e","_kg_hide-input":true,"scrolled":false,"_cell_guid":"da93f329-0d94-40bf-a7c7-1ba111177a69","trusted":true},"cell_type":"code","source":"print(\"Description of Distribuition Saving accounts by Risk:  \")\nprint(pd.crosstab(df_credit[\"Saving accounts\"],df_credit.Risk))\n\nfig, ax = plt.subplots(3,1, figsize=(12,12))\ng = sns.countplot(x=\"Saving accounts\", data=df_credit, palette=\"hls\", \n              ax=ax[0],hue=\"Risk\")\ng.set_title(\"Saving Accounts Count\", fontsize=15)\ng.set_xlabel(\"Saving Accounts type\", fontsize=12)\ng.set_ylabel(\"Count\", fontsize=12)\n\ng1 = sns.violinplot(x=\"Saving accounts\", y=\"Job\", data=df_credit, palette=\"hls\", \n               hue = \"Risk\", ax=ax[1],split=True)\ng1.set_title(\"Saving Accounts by Job\", fontsize=15)\ng1.set_xlabel(\"Savings Accounts type\", fontsize=12)\ng1.set_ylabel(\"Job\", fontsize=12)\n\ng = sns.boxplot(x=\"Saving accounts\", y=\"Credit amount\", data=df_credit, ax=ax[2],\n            hue = \"Risk\",palette=\"hls\")\ng2.set_title(\"Saving Accounts by Credit Amount\", fontsize=15)\ng2.set_xlabel(\"Savings Accounts type\", fontsize=12)\ng2.set_ylabel(\"Credit Amount(US)\", fontsize=12)\n\nplt.subplots_adjust(hspace = 0.4,top = 0.9)\n\nplt.show()\n","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"0a3d67e8cff0c4da83190cd33e2b30d4192eb226","_cell_guid":"e692d92b-024c-4a3a-915a-51eaed3a6ee0"},"cell_type":"markdown","source":"Pretty and interesting distribution..."},{"metadata":{"_uuid":"6328aefb8cf8d1a66dc989559896a8915949e24e","_cell_guid":"dcc175f2-0c3b-44c5-8593-9abcf83ab0d8","trusted":true},"cell_type":"code","source":"print(\"Values describe: \")\nprint(pd.crosstab(df_credit.Purpose, df_credit.Risk))\n\nplt.figure(figsize = (14,12))\n\nplt.subplot(221)\ng = sns.countplot(x=\"Purpose\", data=df_credit, \n              palette=\"hls\", hue = \"Risk\")\ng.set_xticklabels(g.get_xticklabels(),rotation=45)\ng.set_xlabel(\"\", fontsize=12)\ng.set_ylabel(\"Count\", fontsize=12)\ng.set_title(\"Purposes Count\", fontsize=20)\n\nplt.subplot(222)\ng1 = sns.violinplot(x=\"Purpose\", y=\"Age\", data=df_credit, \n                    palette=\"hls\", hue = \"Risk\",split=True)\ng1.set_xticklabels(g1.get_xticklabels(),rotation=45)\ng1.set_xlabel(\"\", fontsize=12)\ng1.set_ylabel(\"Count\", fontsize=12)\ng1.set_title(\"Purposes by Age\", fontsize=20)\n\nplt.subplot(212)\ng2 = sns.boxplot(x=\"Purpose\", y=\"Credit amount\", data=df_credit, \n               palette=\"hls\", hue = \"Risk\")\ng2.set_xlabel(\"Purposes\", fontsize=12)\ng2.set_ylabel(\"Credit Amount\", fontsize=12)\ng2.set_title(\"Credit Amount distribuition by Purposes\", fontsize=20)\n\nplt.subplots_adjust(hspace = 0.6, top = 0.8)\n\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"65b2d7708d9234540d38f9c6f1f7c2f40f527baa","_cell_guid":"202f3cfa-adcb-4903-bc87-c77dbb18c151"},"cell_type":"markdown","source":"Duration of the loans distribuition and density"},{"metadata":{"_uuid":"de0f1619a2b148a438242c4a84b0c3f636b3bbfb","_cell_guid":"d2cf1369-8f55-468f-9f95-4fec12237325","trusted":true},"cell_type":"code","source":"plt.figure(figsize = (12,14))\n\ng= plt.subplot(311)\ng = sns.countplot(x=\"Duration\", data=df_credit, \n              palette=\"hls\",  hue = \"Risk\")\ng.set_xlabel(\"Duration Distribuition\", fontsize=12)\ng.set_ylabel(\"Count\", fontsize=12)\ng.set_title(\"Duration Count\", fontsize=20)\n\ng1 = plt.subplot(312)\ng1 = sns.pointplot(x=\"Duration\", y =\"Credit amount\",data=df_credit,\n                   hue=\"Risk\", palette=\"hls\")\ng1.set_xlabel(\"Duration\", fontsize=12)\ng1.set_ylabel(\"Credit Amount(US)\", fontsize=12)\ng1.set_title(\"Credit Amount distribuition by Duration\", fontsize=20)\n\ng2 = plt.subplot(313)\ng2 = sns.distplot(df_good[\"Duration\"], color='g')\ng2 = sns.distplot(df_bad[\"Duration\"], color='r')\ng2.set_xlabel(\"Duration\", fontsize=12)\ng2.set_ylabel(\"Frequency\", fontsize=12)\ng2.set_title(\"Duration Frequency x good and bad Credit\", fontsize=20)\n\nplt.subplots_adjust(wspace = 0.4, hspace = 0.4,top = 0.9)\n\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"9b6a11a2a266a52520e580c01968ee38a71c33b3","_cell_guid":"0a37eb7a-da07-4b52-a24e-f5c9ad9cfad1"},"cell_type":"markdown","source":"\nInteresting, we can see that the highest duration have the high amounts. <br>\nThe highest density is between [12 ~ 18 ~ 24] months<br>\nIt all make sense.\n"},{"metadata":{"_uuid":"f89cbb713a6a6fcbf8b1aac5b83e749ae437da25","_cell_guid":"575d3bcb-618b-499e-a211-f5eff8df488e"},"cell_type":"markdown","source":"<h2> Checking Account variable </h2>"},{"metadata":{"_uuid":"8eaa1e295c81198428e5e5acf68efc901be3cd3f","_cell_guid":"e825b44c-77b1-4118-bb9a-e4a344545821"},"cell_type":"markdown","source":"First, let's look the distribuition "},{"metadata":{"_kg_hide-output":false,"_kg_hide-input":true,"scrolled":false,"_cell_guid":"7eee1a85-051c-4fd3-a42f-b7c10afed165","_uuid":"f6b6400dccab94229969a881fc805ffacc2dad32","trusted":true},"cell_type":"code","source":"#First plot\ntrace0 = go.Bar(\n    x = df_credit[df_credit[\"Risk\"]== 'good'][\"Checking account\"].value_counts().index.values,\n    y = df_credit[df_credit[\"Risk\"]== 'good'][\"Checking account\"].value_counts().values,\n    name='Good credit Distribuition' \n    \n)\n\n#Second plot\ntrace1 = go.Bar(\n    x = df_credit[df_credit[\"Risk\"]== 'bad'][\"Checking account\"].value_counts().index.values,\n    y = df_credit[df_credit[\"Risk\"]== 'bad'][\"Checking account\"].value_counts().values,\n    name=\"Bad Credit Distribuition\"\n)\n\ndata = [trace0, trace1]\n\nlayout = go.Layout(\n    title='Checking accounts Distribuition',\n    xaxis=dict(title='Checking accounts name'),\n    yaxis=dict(title='Count'),\n    barmode='group'\n)\n\n\nfig = go.Figure(data=data, layout=layout)\n\npy.iplot(fig, filename = 'Age-ba', validate = False)","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"94531d0a22cb1ed2f778cef64055d9f49f5c3666","_cell_guid":"b278620d-be31-46fc-9d1a-f9ac2aea4641"},"cell_type":"markdown","source":"Now, we will verify the values through Checking Accounts"},{"metadata":{"_uuid":"63ded225a79e54f6cef8d7137cfa19ec28bab8ce","_kg_hide-output":false,"_kg_hide-input":true,"_cell_guid":"f4ff7f11-1163-468a-8049-b862e022874b","trusted":true},"cell_type":"code","source":"df_good = df_credit[df_credit[\"Risk\"] == 'good']\ndf_bad = df_credit[df_credit[\"Risk\"] == 'bad']\n\ntrace0 = go.Box(\n    y=df_good[\"Credit amount\"],\n    x=df_good[\"Checking account\"],\n    name='Good credit',\n    marker=dict(\n        color='#3D9970'\n    )\n)\n\ntrace1 = go.Box(\n    y=df_bad['Credit amount'],\n    x=df_bad['Checking account'],\n    name='Bad credit',\n    marker=dict(\n        color='#FF4136'\n    )\n)\n    \ndata = [trace0, trace1]\n\nlayout = go.Layout(\n    yaxis=dict(\n        title='Cheking distribuition'\n    ),\n    boxmode='group'\n)\nfig = go.Figure(data=data, layout=layout)\n\npy.iplot(fig, filename='box-age-cat')","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"289dea7145c6814dbab59cc7e812d62aa5d430ed","_cell_guid":"209ba924-c382-43a4-82d1-6925901dd96f"},"cell_type":"markdown","source":"The old plot that I am trying to substitute with interactive plots"},{"metadata":{"_uuid":"9e59691f08fc14ecbc742798e46c95199f23ed70","_kg_hide-input":true,"_cell_guid":"9744fbaa-07cd-4c74-b782-df20ca9425ab","trusted":true},"cell_type":"code","source":"print(\"Total values of the most missing variable: \")\nprint(df_credit.groupby(\"Checking account\")[\"Checking account\"].count())\n\nplt.figure(figsize = (12,10))\n\ng = plt.subplot(221)\ng = sns.countplot(x=\"Checking account\", data=df_credit, \n              palette=\"hls\", hue=\"Risk\")\ng.set_xlabel(\"Checking Account\", fontsize=12)\ng.set_ylabel(\"Count\", fontsize=12)\ng.set_title(\"Checking Account Counting by Risk\", fontsize=20)\n\ng1 = plt.subplot(222)\ng1 = sns.violinplot(x=\"Checking account\", y=\"Age\", data=df_credit, palette=\"hls\", hue = \"Risk\",split=True)\ng1.set_xlabel(\"Checking Account\", fontsize=12)\ng1.set_ylabel(\"Age\", fontsize=12)\ng1.set_title(\"Age by Checking Account\", fontsize=20)\n\ng2 = plt.subplot(212)\ng2 = sns.boxplot(x=\"Checking account\",y=\"Credit amount\", data=df_credit,hue='Risk',palette=\"hls\")\ng2.set_xlabel(\"Checking Account\", fontsize=12)\ng2.set_ylabel(\"Credit Amount(US)\", fontsize=12)\ng2.set_title(\"Credit Amount by Cheking Account\", fontsize=20)\n\nplt.subplots_adjust(wspace = 0.2, hspace = 0.3, top = 0.9)\n\nplt.show()\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d2879efa0b4e54e202ee51089315f9c2628183cf","_cell_guid":"774c8e58-2eee-4ec5-ae9b-e1432e18c2b9"},"cell_type":"markdown","source":"Crosstab session and anothers to explore our data by another metrics a little deep"},{"metadata":{"_uuid":"98db0c998d53f3a3932b05b89c6efb3c511c63c3","_cell_guid":"d684c524-8bea-4774-904f-c640f9ac78ae","trusted":true},"cell_type":"code","source":"print(pd.crosstab(df_credit.Sex, df_credit.Job))","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"3b1bb68e671a50b90a1e88545619e91dd2cd2390","_cell_guid":"548bfb67-4160-45cd-92bf-df98fd1e013a","trusted":true},"cell_type":"code","source":"plt.figure(figsize = (10,6))\n\ng = sns.violinplot(x=\"Housing\",y=\"Job\",data=df_credit,\n                   hue=\"Risk\", palette=\"hls\",split=True)\ng.set_xlabel(\"Housing\", fontsize=12)\ng.set_ylabel(\"Job\", fontsize=12)\ng.set_title(\"Housing x Job - Dist\", fontsize=20)\n\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"scrolled":false,"_uuid":"594cd815754f297e11b53cbeec840f15f19e6061","_cell_guid":"9adcbb1f-ea69-41f6-b067-e0aaaf249d37","trusted":true},"cell_type":"code","source":"print(pd.crosstab(df_credit[\"Checking account\"],df_credit.Sex))","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d136dca334dba80df467e7b34bf8aa6e777260b3","_cell_guid":"b7c5d494-0fa9-4120-ae70-131615e349eb","trusted":true},"cell_type":"code","source":"date_int = [\"Purpose\", 'Sex']\ncm = sns.light_palette(\"green\", as_cmap=True)\npd.crosstab(df_credit[date_int[0]], df_credit[date_int[1]]).style.background_gradient(cmap = cm)","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"9ddbce3ced63a43cc9c791963354bbd99efa27e5","_cell_guid":"4c1fc98f-4b91-4f8d-a868-71cdd12b5228","trusted":true},"cell_type":"code","source":"date_int = [\"Purpose\", 'Sex']\ncm = sns.light_palette(\"green\", as_cmap=True)\npd.crosstab(df_credit[date_int[0]], df_credit[date_int[1]]).style.background_gradient(cmap = cm)","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"144849d8fe54bdfec7ef341ef3f84b061b3a17a3"},"cell_type":"markdown","source":"## Looking the total of values in each categorical feature"},{"metadata":{"_uuid":"0b46ab68c9c3b8ada44951310ad52587737a59f2","_cell_guid":"6877598e-6b26-44bb-b171-4f6657c0c57b","trusted":true},"cell_type":"code","source":"print(\"Purpose : \",df_credit.Purpose.unique())\nprint(\"Sex : \",df_credit.Sex.unique())\nprint(\"Housing : \",df_credit.Housing.unique())\nprint(\"Saving accounts : \",df_credit['Saving accounts'].unique())\nprint(\"Risk : \",df_credit['Risk'].unique())\nprint(\"Checking account : \",df_credit['Checking account'].unique())\nprint(\"Aget_cat : \",df_credit['Age_cat'].unique())","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"fc4cdbe7701069d2bf7c7309c865fa63374238f8"},"cell_type":"markdown","source":"## Let's do some feature engineering on this values and create variable Dummies of the values"},{"metadata":{"trusted":true,"_uuid":"909b478bf2af28cd21e7f9e90f99848c49007a23"},"cell_type":"code","source":"def one_hot_encoder(df, nan_as_category = False):\n    original_columns = list(df.columns)\n    categorical_columns = [col for col in df.columns if df[col].dtype == 'object']\n    df = pd.get_dummies(df, columns= categorical_columns, dummy_na= nan_as_category, drop_first=True)\n    new_columns = [c for c in df.columns if c not in original_columns]\n    return df, new_columns","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d3ac3838812794fd1b1213732fe5ebe4d22d6775"},"cell_type":"markdown","source":"## Transforming the data into Dummy variables"},{"metadata":{"trusted":true,"_uuid":"bd8cedad464868cff47f03f01a83c902a4a60644"},"cell_type":"code","source":"df_credit['Saving accounts'] = df_credit['Saving accounts'].fillna('no_inf')\ndf_credit['Checking account'] = df_credit['Checking account'].fillna('no_inf')\n\n#Purpose to Dummies Variable\ndf_credit = df_credit.merge(pd.get_dummies(df_credit.Purpose, drop_first=True, prefix='Purpose'), left_index=True, right_index=True)\n#Sex feature in dummies\ndf_credit = df_credit.merge(pd.get_dummies(df_credit.Sex, drop_first=True, prefix='Sex'), left_index=True, right_index=True)\n# Housing get dummies\ndf_credit = df_credit.merge(pd.get_dummies(df_credit.Housing, drop_first=True, prefix='Housing'), left_index=True, right_index=True)\n# Housing get Saving Accounts\ndf_credit = df_credit.merge(pd.get_dummies(df_credit[\"Saving accounts\"], drop_first=True, prefix='Savings'), left_index=True, right_index=True)\n# Housing get Risk\ndf_credit = df_credit.merge(pd.get_dummies(df_credit.Risk, prefix='Risk'), left_index=True, right_index=True)\n# Housing get Checking Account\ndf_credit = df_credit.merge(pd.get_dummies(df_credit[\"Checking account\"], drop_first=True, prefix='Check'), left_index=True, right_index=True)\n# Housing get Age categorical\ndf_credit = df_credit.merge(pd.get_dummies(df_credit[\"Age_cat\"], drop_first=True, prefix='Age_cat'), left_index=True, right_index=True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"collapsed":true,"_uuid":"35deba00af149bdeee83e89db70464e735c3bdff"},"cell_type":"markdown","source":"## Deleting the old features"},{"metadata":{"trusted":true,"_uuid":"3a7d9792cbcb54c6aa4f9f6628303762696d426e"},"cell_type":"code","source":"#Excluding the missing columns\ndel df_credit[\"Saving accounts\"]\ndel df_credit[\"Checking account\"]\ndel df_credit[\"Purpose\"]\ndel df_credit[\"Sex\"]\ndel df_credit[\"Housing\"]\ndel df_credit[\"Age_cat\"]\ndel df_credit[\"Risk\"]\ndel df_credit['Risk_good']","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"338474de1adaa3ff6075f0eef44cb3eec08bd9a3"},"cell_type":"markdown","source":""},{"metadata":{"_uuid":"d03a1fb1cc1140ca0a094bd3746dae62ebfa18d9","_cell_guid":"6513753a-97cb-4b60-b199-87611f7dd411"},"cell_type":"markdown","source":"# **5. Correlation:** <a id=\"Correlation\"></a> <br>\n- Looking the data correlation\n<h1>Looking the correlation of the data"},{"metadata":{"_uuid":"11e60e581715f44cc87edbee7ffec30fe5d3b7da","_cell_guid":"ed4c5ac8-ac85-44d8-8ba3-f6d3fec8a151","trusted":true},"cell_type":"code","source":"plt.figure(figsize=(14,12))\nsns.heatmap(df_credit.astype(float).corr(),linewidths=0.1,vmax=1.0, \n            square=True,  linecolor='white', annot=True)\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"783e1c6f2d23d827800df319ce41c9bc200c0e67","_cell_guid":"f98f6b74-20d1-4b32-a549-14b0079fd4e1"},"cell_type":"markdown","source":"# **6. Preprocessing:** <a id=\"Preprocessing\"></a> <br>\n- Importing ML librarys\n- Setting X and y variables to the prediction\n- Splitting Data\n"},{"metadata":{"_uuid":"32b245afaf3c349eedddfe0c106b24eede56af60","_cell_guid":"7d5906f7-bebd-49e1-a1a5-9db82ae75e16","trusted":true},"cell_type":"code","source":"from sklearn.model_selection import train_test_split, KFold, cross_val_score # to split the data\nfrom sklearn.metrics import accuracy_score, confusion_matrix, classification_report, fbeta_score #To evaluate our model\n\nfrom sklearn.model_selection import GridSearchCV\n\n# Algorithmns models to be compared\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.tree import DecisionTreeClassifier\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.discriminant_analysis import LinearDiscriminantAnalysis\nfrom sklearn.naive_bayes import GaussianNB\nfrom sklearn.svm import SVC\nfrom xgboost import XGBClassifier\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"34718c735d0ddbd8ad676e34cba8950755998809"},"cell_type":"code","source":"df_credit['Credit amount'] = np.log(df_credit['Credit amount'])","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"b0eb1cac1b343143c7ff5ccff82931768167c2de","_cell_guid":"be683db1-dc4f-49f3-86fa-e17eb90c3aef","trusted":true},"cell_type":"code","source":"#Creating the X and y variables\nX = df_credit.drop('Risk_bad', 1).values\ny = df_credit[\"Risk_bad\"].values\n\n# Spliting X and y into train and test version\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.25, random_state=42)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"18d0c87ea7bfdbec2cb86e5517a026f3f0657570"},"cell_type":"code","source":"# to feed the random state\nseed = 7\n\n# prepare models\nmodels = []\nmodels.append(('LR', LogisticRegression()))\nmodels.append(('LDA', LinearDiscriminantAnalysis()))\nmodels.append(('KNN', KNeighborsClassifier()))\nmodels.append(('CART', DecisionTreeClassifier()))\nmodels.append(('NB', GaussianNB()))\nmodels.append(('RF', RandomForestClassifier()))\nmodels.append(('SVM', SVC(gamma='auto')))\nmodels.append(('XGB', XGBClassifier()))\n\n# evaluate each model in turn\nresults = []\nnames = []\nscoring = 'recall'\n\nfor name, model in models:\n        kfold = KFold(n_splits=10, random_state=seed)\n        cv_results = cross_val_score(model, X_train, y_train, cv=kfold, scoring=scoring)\n        results.append(cv_results)\n        names.append(name)\n        msg = \"%s: %f (%f)\" % (name, cv_results.mean(), cv_results.std())\n        print(msg)\n        \n# boxplot algorithm comparison\nfig = plt.figure(figsize=(11,6))\nfig.suptitle('Algorithm Comparison')\nax = fig.add_subplot(111)\nplt.boxplot(results)\nax.set_xticklabels(names)\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"c3ec3f072212092709c00591d203b5ba674d8ec5"},"cell_type":"markdown","source":"Very interesting. Almost all models shows a low value to recall. \n\nWe can observe that our best results was with CART, NB and XGBoost. <br>\nI will implement some models and try to do a simple Tunning on them"},{"metadata":{"_uuid":"55565ef692e7418af62bbd6f23babd7994cd2fc2","_cell_guid":"dc9dca9e-02a8-4fea-8265-eaf631ae39cd"},"cell_type":"markdown","source":"# **7.1 Model 1 :** <a id=\"Modelling 1\"></a> <br>\n- Using Random Forest to predictict the credit score \n- Some of Validation Parameters"},{"metadata":{"_uuid":"69d73aabd9ed02ca77fe5bd29ab943960f0c99ee","_cell_guid":"700ab26d-2e82-4b1c-b575-1f5f5f7a2daf","trusted":true,"_kg_hide-output":true,"scrolled":false},"cell_type":"code","source":"#Seting the Hyper Parameters\nparam_grid = {\"max_depth\": [3,5, 7, 10,None],\n              \"n_estimators\":[3,5,10,25,50,150],\n              \"max_features\": [4,7,15,20]}\n\n#Creating the classifier\nmodel = RandomForestClassifier(random_state=2)\n\ngrid_search = GridSearchCV(model, param_grid=param_grid, cv=5, scoring='recall', verbose=4)\ngrid_search.fit(X_train, y_train)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"420025da50cbf987174da6a4990032a9abc9f2ee"},"cell_type":"code","source":"print(grid_search.best_score_)\nprint(grid_search.best_params_)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"ed2f94c143bd97a4ce0931c6fd82aff9ccc7d434"},"cell_type":"code","source":"rf = RandomForestClassifier(max_depth=None, max_features=10, n_estimators=15, random_state=2)\n\n#trainning with the best params\nrf.fit(X_train, y_train)","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"1708146ad90c695224bff81d73e0a4b37808b3bb","_cell_guid":"df58e526-183d-4136-8bba-181cacd378ff","trusted":true},"cell_type":"code","source":"#Testing the model \n#Predicting using our  model\ny_pred = rf.predict(X_test)\n\n# Verificaar os resultados obtidos\nprint(accuracy_score(y_test,y_pred))\nprint(\"\\n\")\nprint(confusion_matrix(y_test, y_pred))\nprint(\"\\n\")\nprint(fbeta_score(y_test, y_pred, beta=2))","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"18705cc8e5c8a8d7e607a06e6f5b12d40bd5978d","_cell_guid":"b8b0d3d4-0d56-4aba-83f4-d8d2c53c65de"},"cell_type":"markdown","source":"Very sucks results! How can I increase my model? "},{"metadata":{"_uuid":"fe7fc8bb6c6b392a62972132065d971468b85a3d","_cell_guid":"ad8f1404-7812-46a5-9f68-94647930b728"},"cell_type":"markdown","source":"# **7.2 Model 2:** <a id=\"Modelling 2\"></a> <br>"},{"metadata":{"_uuid":"48c5b40e53ad8ed9e6fe5493802445996e597869","_cell_guid":"6e15774a-0b93-41ae-a395-6fde3c1ebd3a","trusted":true},"cell_type":"code","source":"from sklearn.utils import resample\nfrom sklearn.metrics import roc_curve","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d2b3aded7d3373de62dbe57e4c1e1f86870a202c","_cell_guid":"3c980a9e-7cac-4d5b-b0e6-2dde790bb5e4","trusted":true},"cell_type":"code","source":"# Criando o classificador logreg\nGNB = GaussianNB()\n\n# Fitting with train data\nmodel = GNB.fit(X_train, y_train)","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"8cf623e40c3615367b5ed12bc8fdcb5b825ca58d","_cell_guid":"797436e3-1d4b-4748-860d-efb282a1d0ce","trusted":true},"cell_type":"code","source":"# Printing the Training Score\nprint(\"Training score data: \")\nprint(model.score(X_train, y_train))","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"cd3133deabdf47413f070365a4da6b5566e19e7f","_cell_guid":"9649fb32-39e0-495a-8a7b-dd659bb24dc9","trusted":true},"cell_type":"code","source":"y_pred = model.predict(X_test)\n\nprint(accuracy_score(y_test,y_pred))\nprint(\"\\n\")\nprint(confusion_matrix(y_test, y_pred))\nprint(\"\\n\")\nprint(classification_report(y_test, y_pred))","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"82d1a0bc17bc3c2a7958beb8a3c6af8295b1c8ea"},"cell_type":"markdown","source":"With the Gaussian Model we got a best recall. "},{"metadata":{"_uuid":"08ebd2c7ba98564dc75716480af7291b5160e222"},"cell_type":"markdown","source":"## Let's verify the ROC curve"},{"metadata":{"_uuid":"d5e668be55353b5129a63ebd524ed83e6a439de4","_cell_guid":"8e3f97c0-7df5-4386-a248-5fffcc7dbc3b","trusted":true},"cell_type":"code","source":"#Predicting proba\ny_pred_prob = model.predict_proba(X_test)[:,1]\n\n# Generate ROC curve values: fpr, tpr, thresholds\nfpr, tpr, thresholds = roc_curve(y_test, y_pred_prob)\n\n# Plot ROC curve\nplt.plot([0, 1], [0, 1], 'k--')\nplt.plot(fpr, tpr)\nplt.xlabel('False Positive Rate')\nplt.ylabel('True Positive Rate')\nplt.title('ROC Curve')\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"3dd2ac7c21fa16ee53db73a517255544c3618d30","_cell_guid":"06356183-c43a-4258-9fda-557b16fda15e","trusted":true},"cell_type":"code","source":"from sklearn.model_selection import KFold\nfrom sklearn.model_selection import cross_val_score\nfrom sklearn.pipeline import Pipeline\nfrom sklearn.pipeline import FeatureUnion\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.decomposition import PCA\nfrom sklearn.feature_selection import SelectKBest","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"a2c6217b39666c7ab8bd117d452dac5ac459b49b"},"cell_type":"code","source":"features = []\nfeatures.append(('pca', PCA(n_components=2)))\nfeatures.append(('select_best', SelectKBest(k=6)))\nfeature_union = FeatureUnion(features)\n# create pipeline\nestimators = []\nestimators.append(('feature_union', feature_union))\nestimators.append(('logistic', GaussianNB()))\nmodel = Pipeline(estimators)\n# evaluate pipeline\nseed = 7\nkfold = KFold(n_splits=10, random_state=seed)\nresults = cross_val_score(model, X_train, y_train, cv=kfold)\nprint(results.mean())","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"6a258a8b79778264cbbd2f0e290df3b936c00a65"},"cell_type":"code","source":"model.fit(X_train, y_train)\ny_pred = model.predict(X_test)\n\nprint(accuracy_score(y_test,y_pred))\nprint(\"\\n\")\nprint(confusion_matrix(y_test, y_pred))\nprint(\"\\n\")\nprint(fbeta_score(y_test, y_pred, beta=2))","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"ac6b16a05b55bc3f93eece77a2c498c74b7203b5"},"cell_type":"markdown","source":""},{"metadata":{"_uuid":"f14be7e8c492b371c1499ea8a420f438669c2910"},"cell_type":"markdown","source":"## Implementing a pipeline of models "},{"metadata":{"trusted":true,"_uuid":"7dc4f8c1d2c1ee7360999f0be703b68322b49b8f","scrolled":false,"_kg_hide-output":true},"cell_type":"code","source":"#Seting the Hyper Parameters\nparam_test1 = {\n 'max_depth':[3,5,6,10],\n 'min_child_weight':[3,5,10],\n 'gamma':[0.0, 0.1, 0.2, 0.3, 0.4],\n# 'reg_alpha':[1e-5, 1e-2, 0.1, 1, 10],\n 'subsample':[i/100.0 for i in range(75,90,5)],\n 'colsample_bytree':[i/100.0 for i in range(75,90,5)]\n}\n\n#Creating the classifier\nmodel_xg = XGBClassifier(random_state=2)\n\ngrid_search = GridSearchCV(model_xg, param_grid=param_test1, cv=5, scoring='recall')\ngrid_search.fit(X_train, y_train)","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"3681e7b5bea9e6582d3cedc31a93928be4663a49"},"cell_type":"markdown","source":"How can I avoid of this "},{"metadata":{"trusted":true,"collapsed":true,"_uuid":"d2a3bdd83f3fbdc7b0221e9a719768e8044f1536","_kg_hide-input":true},"cell_type":"markdown","source":"# Please, how can I increase this sucks models ?"},{"metadata":{"trusted":true,"_uuid":"a47030a77584196bd9c806ee5970677a1eb23f2a"},"cell_type":"code","source":"grid_search.best_score_\ngrid_search.best_params_","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"cc0352311ad308a85f1ecef587df26e82f76173d"},"cell_type":"code","source":"y_pred = grid_search.predict(X_test)\n\n# Verificaar os resultados obtidos\nprint(accuracy_score(y_test,y_pred))\nprint(\"\\n\")\nprint(confusion_matrix(y_test, y_pred))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"04c51fdb265f8f00d20c3c4e70e2924765bd469e"},"cell_type":"markdown","source":"How can I avoid this error? \n\n/opt/conda/lib/python3.6/site-packages/sklearn/preprocessing/label.py:151: DeprecationWarning:\n\nThe truth value of an empty array is ambiguous. Returning False, but in future this will result in an error. Use `array.size > 0` to check that an array is not empty.\n"},{"metadata":{"trusted":true,"_uuid":"3a6c1313a9f649d9f151acd4ee82bf9168184343"},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"name":"python","version":"3.6.6","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":1}