I used R to create a linear regression model with a single variable and multivariable analysis for an Ecommerce Customer Device Usage Data Set. The following visuals were done in the "ggplot2" library. The model was able to successfully make strong predictions on future user data and behaviour and prvovide valuable insights.
AVERAGE SESSION LENGTH VS YEARLY AMOUNT SPENT
TIME ON WEBSITE VS YEARLY AMOUNT SPENT
PAIR PLOT OF ALL CONTINUOUS VARIABLES
RESIDUALS ANALYSIS
EVALUATING THE MULTIPLE REGRESSION MODEL FINDINGS
By using a multiple linear model, we have created a much more accurate predictor of the response variable.
- R2 went from 0.65 to 0.98
- Mean Absolute Percentage Error 0.07 to 0.01
- The Root Mean Square Errorwent from 47.14 to 9.97 dollars.
To get better insight and understanding on the Regression Model, I suggest checking Main.R for the source code. All my code is properly documented throughout my project.