Linear Regression is a traditional supervised Machine Learning technique commonly used for prediction. The approach is frequently used in many applications involving sequencing or forecasts for the time-series domain. The efficiency of the model depends on the characteristics of the data and some other external factors. One notable limitation of this model is its reliance on linear data. When presented with non-linear data, its performance can be severely impacted. To address this issue, researchers have suggested an alternative approach known as piecewise linear regression. Piecewise linear regression is a variant of linear regression that involves partitioning the data points based on their linear characteristics, followed by the application of linear regression within each partition. The primary difficulty associated with piecewise linear regression is identifying the optimal approach for partitioning the data, which can be thought of as locating the breakpoints. The main objective of this paper is to introduce a strategy for partitioning and identifying breakpoints based on the correlation properties of the data points. The effectiveness of this approach was evaluated using open-access COVID-19 data in the context of a prediction model. The findings indicate that the proposed partitioning and breakpoint identification method outperforms other existing systems, suggesting that it is a viable option that is relatively close to the optimal solution.