Comparative Study of Proposed linear regression algorithm to Scikit-Learn Algorithm

The majority of real-world applications of machine learning employ supervised learning. With an input variable (x) and an outcome variable (y), supervised learning allows one to apply an algorithm to train the mapping function from the input to the result, which is Y = f(X). The goal is to approximate the mapping function as nearly as feasible to predict the output variables (Y) with fresh input data (x). Techniques used by supervised machine learning algorithms include support vector machines, logistic and linear regression, decision trees, and multi-class classification. To enable supervised learning, the data used to train the system should be labelled with the appropriate results. Often, linear regression is used to model the relationship that exists between a dependent variable and one or more independent variables. Using estimated parameters from the data, linear models are created. By trying to fit a predictive model to a collection of observed data values, linear regression can be used to forecast and predict outcomes. To minimize error, the least-squares method is widely used to fit linear regression models. Problems with classification and regression fall within the heading of supervised learning. The goal of both issues is to develop a compact model that can precisely predict the value of the dependent variable from the attribute variables. The dependent attribute is different for the two tasks since it is categorical for classification and numerical for regression. Using a dataset, we are proposing an algorithm for linear regression in this study. We are also comparing their implementations with and without Sklearn.