PREDICTING THE TOP THREE FINISHERS OF THE 2016 KENTUCKY DERBY USING LOGISTIC REGRESSION

There are many methods of predicting or forecasting within the realm of multivariate statistical analysis. Different methods are suitable for different purposes.  The aim of this study was to apply logistic regression... [ view full abstract ]

There are many methods of predicting or forecasting within the realm of multivariate statistical analysis. Different methods are suitable for different purposes.  The aim of this study was to apply logistic regression analysis to effectively build a model that predicts whether a given horse might finish in the top three places of a race.

Logistic Regression modeling is one of several multivariate tools available to analyze complex data in order to find patterns and relationships that may not be apparent otherwise.  The power of this method is the ability to predict outcomes or classifications of future events or observations.  Logistic regression works with continuous and/or discrete predictor variables to find the probability of group membership for the response variable based on predictor variables.

In this project we were interested in whether or not a particular horse in the Kentucky Derby would finish in the top 3. In building our data set we limited our data to only the information that would be available pre-race to the average individual.  By using logistic regression we were able to create a model that incorporated six different independent variables.  The model correctly produced the highest probabilities for the top 3 finishers for the 2016 Kentucky Derby. In further testing we examined the difference in prediction accuracy between logistic regression and decision tree models for both the Kentucky Derby and the Preakness Stakes.