Behavioral Segmentation with Product Estimation using K-Means Clustering and Seasonal ARIMA

Definite purchasing patterns, usage, interests, engagement level, and responses emphasize the consumer's needs and want. Relating to these behavioral patterns, supermarkets can generate the right experience for their buyers, thus accounting for their interaction with the companies in real-time. The proposed methodology will allow the supermarket and associated brands to have an even better qualitative customer-centric approach, a key outcome of which is an increase in their CLV (or Customer Lifetime Value). The framework consists of an unsupervised Machine Learning technique of K-means clustering applied on a supermarket sales dataset. The resultant clusters will depict distinct segments of customers grouped on behavioral patterns extending the conventional RFM analysis approach. Further, the Seasonal ARIMA method models the time-series data to estimate the sales of top-selling products. It is a suggested approach to maintain and manage the supermarket-base more effectively along with a satisfied customer base.