8.Sales Forecast Model - Linear Regression.Rmd

---
title: "Sales Forecast Model - Linear Regression"
output: html_notebook
---


```{r Load packages}
library(tidyverse)
library(openxlsx)
```


```{r Load file}
oreos <- read.xlsx("~/Documents/GitHub/Machine-Learning-R/Machine-Learning-R/datasets/Oreos.xlsx", skipEmptyRows = TRUE)
```


```{r How does shelf position influence Oreos sale}
head(oreos)
```



```{r Build Linear Regression Model}

# Sales as explained by height in feet; must specify dataset
oreo_model_lm <-  lm(Sales ~ factor(Height.in.feet), data = oreos) 

```



```{r Summarize regression model}

#Print model summary
oreo_model_lm %>% summary()

```


A positive coefficient indicates that as the value of the independent variable increases (height), the mean of the dependent variable (sales) also tends to increase.

At 5 feet sales are 28. Now we interpret 6 and 7 feet in relation to the 5 feet shelf location.

At 6 feet sales increase by 34 units, but at 7 feet sales are only increased by 15 units in relation 5 feet. Overall, 6 feet is better location for our Oreos.

p-values are <.05 for both independent variables


How do we write our equation if we want to use this linear model to forecast sales?

Sales = 28.5 + 34*(6 feet) + 15.75*(7 feet)

Luckily, for us we can use R to do the math for us.

```{r Create new dataframe for our forecast}

  explan_oreos <- tibble(oreos)
  
  head(explan_oreos)
  
```

```{r Predict oreo sales}

#Call predict - vector of predictions
predict(oreo_model_lm, explan_oreos)

```


```{r Create dataframe with our predictions}

#Put predictions inside a data frame
prediction_oreo <- explan_oreos  %>% 
  mutate(forecast_sales = predict(oreo_model_lm, explan_oreos))

#Add residuals as a column
prediction_oreo <- prediction_oreo %>% 
  mutate(errors = forecast_sales - Sales)

#check results
head(prediction_oreo)
```



```{r Summarize sales data on our forecast}
prediction_oreo %>% 
  group_by(Height.in.feet) %>% 
  summarize(avg_sales = mean(forecast_sales),
            sum_sales = sum(forecast_sales))
```