Predicting NBA Players' Salaries

Description

Predicting NBA Players' Salaries is a project in which machine learning models were made to predict NBA salaries for the 2021-2022 NBA season and determine if players were overpaid or underpaid. In order to do so, the first step was to load data. 2 different datasets were used. NBA players stats for the 2021-2022 season that were used can be found here and their salaries from the same season can be found here. The data had to be cleaned before it could be merged, aligning each player's statistics with their salary. Next, important features had to be identified. To do so, the correlation between each potential feature and salary was determined, and features with a high enough correlation were used. Instead of predicting a player's exact salary, salaries were split into 'bins' of size 5,000,000 from 0 to 50,000,000 (this range encompassed all of the salaries) and these were calculated in the models. The data had to be split into test and training data before models could be created. After training and making predictions with a K-Nearest Neighbors Classifier Model, a Random Forest Classifier Model, and a Support Vector Machine Model, the hyperparameters were tuned for each of them. The best model was a Random Forest Classifier Model, which had an accuracy of 73%.