This project aims to provide personalized recommendations for users based on their past behavior and preferences. The model also includes analytical capabilities that allow for the exploration of user behavior and product trends, that can be pinpointed down to the single user depending on factors such as location, time, etc.
The proposed Python model bifurcates the consumer pool into sections based on similarities in purchase patterns. For this, we use -
- K-Means
- Agglomerative clustering
- Apriori model
Based on these bifurcations, we can predict and suggest the possible set of products that the consumer can purchase.
The dataset used in this project can be accessed from https://archive.ics.uci.edu/dataset/352/online+retail It is a transnational data set that contains all the transactions occurring between 01/12/2010 and 09/12/2011 for a UK-based and registered non-store online retail.
Link to [models.ipynb]