Customer Prediction and Segmentation through Data Mining and Machine Learning to Recommend Business Strategies for a Small Law Firm

A local law firm was interested in obtaining value from the data they had collected, and were familiar with data science in general, but lacked the expertise to apply these concepts.  The purpose of this project was to provide the firm with actionable insights from proprietary data.  The data was analyzed to determine potential value from implementing data science into business marketing and office expansion.   

This project had three main goals. The first was to provide the firm with customer segmentation data relating to whether potential customers can be separated into different clusters. PCA and K-Means clustering were used to identify two distinct customer segments. The second was to develop machine learning algorithm(s) that best predict whether a potential customer will be profitable. Logistic regression and random forest models provided example use cases for customer classification for small businesses. The final project objective was to provide the firm with a system to identify zip codes in other markets with similar demographics to current customers. K-nearest neighbors (k-NN) and United States census data were used to calculate the similarity scores between zip codes to provide a list of potential zip codes for expansion. 

The results of this project will help the business to optimize their marketing strategy, which will allow for more efficient marketing and increase company revenue. In addition, they will help the business with office growth and entrance into new markets.