An Accurate and Fast Computational Python Based Module for Linear Regression Analysis in Data Science Applications

Due to the increase in computational power of digital electronics in past decades, Data Science has become an emerging field and attracted much research interest for development in various sectors. The usage of Python programming language is tremendously growing due to the large community and its ease of use and its support with powerful scripting algorithms for many engineering applications. One of the simple and basic technique in Data Science is Linear Regression. Traditional Libraries like ‘scikit-learn’ and ‘SciPy’ are built on top of other libraries like ‘NumPy’ and ‘pandas’. In this study, a python module is created for performing simple linear regression named “LinReg” using only python's built-in functions and data types. This allows faster computation of the regression coefficients and the evaluation metrics like slope, intercept, and mean absolute error. A test is also conducted to compare the computational time of the module with the above mentioned two libraries. However, the analogies are given to compare and pros and cons of this work and its future scope is discussed on such modules for various other algorithms in Data Science.