Implementing a Linear Regression Gradient Descent Model for Admission Process Framework

Abstract

Abstract: Machine Learning (ML) is the fundamental learning paradigm in the scientific community having a wide range of applications in vivid domains. The

hidden underlying patterns in data can be easily identified with use of popular ML algorithms. The meaningful pattern provides insight information extracted from the data.   In so doing, human incapability hinders the process of recognising meaningful patterns in the given data sets. Such fine exemplary thoughts are given to machines with suitable algorithms and it can detect not only the finer patterns, but also provides meaningfulness of data spread in the domain. The area of ML is a blend of mathematics, probability, statistics and allied sciences in an articulated way and thus endows the ability to “learn and adapt”. A generalized Gradient Descent (GD) based model is proposed which can be implemented on any dataset. The model is tested to forecast a student’s admission on his (her) GRE score. Proposed model reflects good accuracy and the Pearson Correlation coefficient suggests the pertinent relationship among different attributes. The model also focuses the underlying mathematical derivation to a minimum to comprehend.

Index Terms: Linear Regression, Gradient Descent, RMS Error, MSE, ERM, Pearson Correlation