This project implements the Segmented Linear Regression Model (SLRM), an alternative to traditional Artificial Neural Networks (ANNs). The SLRM models datasets with piecewise linear functions, using a neural compression process to reduce complexity without compromising precision beyond a user-defined tolerance. The core of the solution is the compression algorithm, which transforms an unordered dataset (DataFrame / X, Y) into a final, highly optimized dictionary, ready for prediction.