Performance Enhancement Of Customer Segmentation Using A Distributed Python Framework, Ray

Over the years, there has been a huge popularity of the recommender systems worldwide. Recommender systems have been implemented over several domains ranging from recommendations for videos and movies to that for products and applications, and many more. The algorithms, which are used for recommender systems, implement segmentation of the customer based on several attributes. These algorithms are time-consuming and require comparatively high computation power. This work deals with the parallelization of different algorithms for simple customer segmentation in the Python environment using the framework, Ray. The dataset for this work includes a huge list of purchases that are carried out by 4000 customers, over a year. The parallelization is carried out throughout the multicores of CPU and the cores of GPU. Additionally, the work also shows the speedup that is obtained after parallelization, for analyzing the overall increase in performance.