Analyzing Grammy, Emmy, and Academy Awards Data Using Regression and Maximum Information Coefficient

Prediction of winners has always been fascinating to many people, whether it be for sports, lottery, presidential election, or performing arts awards. The basic approach for prediction is to build a model from the observed data with known outcome labels, and use the model to determine the outcomes of the new observations. It is a typical classification problem in data analysis. An important aspect in the model building process is to select attributes (independent variables) of the data that have the most discriminating power for classification. In this paper, we present our study using multiple logistic regression and multiple linear regression modeling along with Maximal Information-based Nonparametric Exploration (MINE) statistics to analyze three of the most well-regarded awards in the entertainment industry - the Oscars, Emmys, and Grammys that are high-profile awards given annually to top artists in the areas of film, television, and music.