CORRELATION MATRIX CALCULATOR USING NUMPY

Abstract

Exploratory Data Analysis (EDA) plays a critical role in understanding relationships between variables in a dataset before applying machine learning or statistical models. One of the most widely used techniques in EDA is correlation analysis, which quantifies the strength and direction of relationships between numerical variables. This project presents the development of a lightweight and interactive correlation matrix analysis tool implemented in Python using the NumPy numerical computing library and visualization capabilities of Matplotlib. The system enables users to upload custom CSV datasets and automatically computes the correlation matrix using a manual implementation of the Pearson correlation coefficient. The application performs mean normalization, covariance computation, and standard deviation scaling to produce a complete correlation matrix representing pairwise relationships among features. The computed matrix is then visualized through a heatmap representation, allowing users to easily identify patterns and dependencies within the dataset. In addition to visualization, the tool performs automated analysis to detect highly correlated feature pairs based on a configurable threshold. This functionality assists in identifying redundant variables and potential multicollinearity issues, which are important considerations during feature selection and model preparation in machine learning workflows.