Title:
NBA HOME WINS PREDICTION

Authors:
Langxi Wang

ABSTRACT:
Aim: This study aimed to build a predictive model for NBA Home Wins usinglogistic regression.
Method: A public data was used in this study. All the records were randomly assigned into 2 groups: training sample (50%) and testing sample (50%). Logistic regression was built using
training sample: artificial neural network and linear regression.
Results: Home teams won 1474 of 2230 games, or 66.1%. According to the logistic regression, FG_PCT_home, FT_PCT_home, FG3_PCT_home, AST_home, REB_home, FG_PCT_away, FT_PCT_away, FG3_PCT_away, AST_away, REB_away, sum_PF_home, sum_PF_guest were significant predictors of the winning of the home teams. The area under curve was 0.9815. The optional cutoff time is 0.52. The mis-classification error was 0.06. The sensitivity rate is about 95.4% and the specificity is 90.4%.
Conclusions: In this study, we identified important of predictors of NBA Home Wins in the United States, for example,% of made shots and total of personal fouls. This tool will be very helpful to understand features determining NBA Home Wins and to maximize the likelihood to win.
Literature review: Before researching my topic, I also referred to many other similar experiments. These included Fadi Thabtah, Li Zhang & Neda Abdelhamid on the use of machine and artificial intelligence to learn and analyze historical data to predict the likelihood of a team's victory. To achieve these goals, the article selects several machine learning methods that use different learning schemes to derive models, including Naive Bayes, artificial neural networks, and decision trees, among others. Several learning schemes were used to finally arrive at which NBA team would win.