Cluster analysis of high-dimensional customer data from a subscription-based business

Cluster analyses are an established method for identifying natural groupings of customers for customer segmentation. However, the unsupervised nature of clustering algorithms and the high-dimensionality of customer data complicate the analysis at all stages. This project presents the results from a cluster analysis of high-dimensional customer data from a subscription-based software company. The analysis tested multiple dimensionality reduction methods, outlier and noise detection methods, and clustering algorithms (including deep neural networks). The results and models from the analysis can be used to inform strategy around customer support and feedback, and can serve as the basis from which additional analyses can be conducted.Master of Science in Information Scienc