Customer Segmentation and Resource Allocation Using Machine Learning Methods

This study delves into the effectiveness of clustering methods for segmenting a less responsive portion of a customer database. Utilizing a dataset of 270,000 customers, we applied a k-means algorithm to divide the dataset into seven distinct groups. Through an in-depth analysis of the centroids of each cluster, two segments characterized by a lower propensity for the desired purchasing pattern were identified. We further employed neural networks to model customer behavior within these segments, aiming to identify individuals least likely to make future purchases. This approach enabled us to assign probabilities of future orders to each customer, optimizing our catalog mailing strategy. By focusing our marketing efforts on customers with a higher likelihood of ordering, we reallocated over $200,000 from catalog mailings to more active customer segments. This research showcases an effective segmentation strategy and demonstrates the utility of neural networks in predicting customer behavior, enhancing the precision of targeted marketing campaigns.