top of page

Online shop recommendation system

Writer's picture: Philip LaiPhilip Lai

Project Objective

Based on the store's sales records, build a recommendation system to predict the next potential item that may be purchased.


Data Ingestion

The data is provided by Kaggle.


Data Processing

1. Create a profit table.

2. Extract several product-combinations based on sales records, according to the order.

3. Set the min_support, min_lift, max_length, and train the Apriori algorithm using the list of combinations to obtain the confidence, support and lift of each combination.

4. Merge the profit table with the analysis results from Apriori, and rank the recommended list based on the expected profit of product combinations.

5. Finally, present the result using a sunburst chart.




Analysis Methods

Apriori is an algorithm based on Bayes' theorem, which utilizes sales records to infer the support and confidence of product combinations.


Presentation

Output a list of corresponding products that have a positive correlation with the certain product.



2 views0 comments

Recent Posts

See All

Comments


bottom of page