conversion-influencers

Goal:

Identify work opportunities that will have the greatest positive impact on customer purchase rate

What:

predictive model to output a list of features that influence whether or not a searching customer decides to purchase a product, ordered by importance

Methodolgy:

• Clickstream data, feature creation and selection
• Random Forest model, feature importance
• xgBoost model, feature importance

Results:

Product inventory and Customer geography identified as 2 important features and two bodies of work completed as a result

Note:

Code file is my working draft and contains everything I trialed; not in the cleanest state with the optimal amount of comments; please forgive 🙏