Research On Prediction Method Of NBA Data Visualization And Integrated Learning Model

Data-driven event analysis has gradually become the backbone of modern competitive sports analysis.Increasingly,competitive sports data analysis tasks use computer vision and machine learning models for intelligent data analysis.This direction aims to study how to use the existing computer vision technology and machine learning model,combined with visual analysis and human-computer interaction technology to help solve the tasks of data collection,data cleaning and data mining in sports data analysis,so as to improve the efficiency and accuracy of task completion.Existing basketball match analysis systems focus on analyzing player and team data.The existing basketball game analysis system pay no attention to the analysis of the team’s season winning data and game time series data.Existing basketball game analysis system ignores All-Star player predictions.Therefore,this thesis against for the regular statistical data of NBA players and teams and the time series data of NBA games.This thesis designs an interactive visualization system NPIPVis.In addition,an ensemble learning-based NBA All-Star player prediction model SRR-Voting is proposed.This thesis mainly includes the following research contents:1.In the task of NBA data visualization,this article first uses D3.JS draws a number of basic interactive visualization components,such as heat maps and radar maps,and then uses PAOHvis to draw a dynamic hypergraph of team season wins and losses according to the 1980-2022 NBA team’s regular statistical data.At the same time,it uses Calliope to draw NBA visual data stories according to the 1980-2022 NBA player’s regular statistical data.Finally,it uses i Storyline narrative visualization technology to draw a game plot narrative visualization component according to the 1980-2022 NBA game timing data,Help ordinary audiences to provide rich visual charts,so that ordinary audiences can better understand the trend of NBA players and teams and the game situation.2.In the prediction task,this thesis uses the 1980-2022 regular stats of the player’s team as a basis.This thesis proposes an all-star player prediction model SRR-Voting based on ensemble learning.The SRR-voting starts from the existing minority and majority samples,was proposed using synthetic minority oversampling technique and Random Under Sampler methods to generate and eliminate samples of a certain size to balance the number of all-star and average players in the datasets.A random forest algorithm was introduced next to extract and construct the features of players and combined with the voting integrated model to predict the all-star players,using Grid Search CV,to optimize the hyperparameters of each model in integrated learning and then combined with K-Fold cross-validation to improve the generalization ability of the model,and,finally,the SHAP model was introduced to enhance the interpretability of the model.The experimental results of comparing the SRR-voting model with six common models show that accuracy and F1-score and recall metrics are significantly improved,which verifies the effectiveness and practicality of the SRR-voting model.This article combines data visualization and machine learning to design a National Basketball Association data visualization system to help the general audience visualize game data and predict all-star players which can be extended to other sports events or related fields.