Website Cookie Data
Business Objective: E-commerce website is interested in improving its transaction rate to drive further profit. How can we predict that a visitor will transact on the website?
Dataset: Website Cookie Data with 16615 observations and 21 variables
​
Understanding the transaction rate of customers to drive further profit on an e-commerce website.
Approach:
Target variable was Transaction revenue (when greater than 0, it indicates that a customer purchased something and generated revenue)
For all numeric variables used a normalization scaling method
Training/testing split of the data: 80/20
Ran about 7 models – Logit regression, Decisions tress, Gradient Boosting, Random Forest, Support Vector Machine and Neural networks (ANN)
Compared the specificity and sensitivity of each of these models
​
Result:
Decision tree had an Accuracy of 89.9% while its Specificity was 94.9% and Sensitivity was 59.5%. Neural Networks had an Accuracy of 89.25% while its Sensitivity was 97.2% (the highest) and the Specificity was 87.9%.
​
​
INSIGHTS
The key business objective is to find ways to drive further revenue generation, my insights are:
By optimizing the engagement across the entire website and driving higher number of page views per visit, this drives higher conversions
Traffic from Google is also much higher quality than other search sites – we should focus the marketing dollars onto Google Certain days of the month also show potentially a higher intend to shop – however, this seems a bit more random and harder to capitalize
Organic vs. inorganic traffic didn’t seem to be a key driver of customer behavior
These three findings can help guide in terms of where to allocate the most of amount of resources and optimize long-term return on the investment – whether it’s a few million dollars to upgrade our website or additional marketing spend to drive interim traffic.