This is an e-commerce data set examining different variables related to Amazon.ca browsing sessions and purchases. The response variable, Revenue, indicates whether or a purchase and sale were made. Some of the predictor variables include: Month, SpecialDay (whether or not the purchase was made near a holiday), and BounceRates (visitors who visit a page and then leave, with no purchase or other action performed). Other variables describe visitor characteristics, geography, Google Analytics, etc.
The data set contains 12,330 observations of online shopping sessions. From these sessions, only 1,908 involved a purchase and sale, while 10,422 sessions did not. This is a sale rate of about 15.5%. The response variable Revenue indicates whether or not an online shopping session resulted in a sale or not. Using non-linear Support Vector Machine (SVM) and random forest modelling, this report will attempt to predict the outcome of sale, based on the combinations of the predictor variables for each Amazon.ca browsing session.