dfd and erd

2.1.1Background:

The NBA is one of the major leagues with the highest salary for American athletes. Since each team has a salary cap, it is very important for the team manager to determine the player's salary more effectively (to avoid premium contracts). The researchers found that in terms of paying for NBA players, play, player experience (league years), assists, rebounds and fouls are all statistically significant factors, but we also found 3-pointers and PER values ​​to be insignificant. Variables, we perform a stepwise regression backward, eliminating irrelevant independent variables one at a time (the least important each time) until the model contains only important variables. Similarly, even the statistical data of the stepwise model has been the original model, but the same variable has statistical significance.

2.1.2Factors taken into consideration:

The determination of the player's salary is a very complicated matter. It is difficult to take all factors into consideration. The factors affecting them mainly include:

1. Player factors Player's skill level: focus, defensive player's off-site factors: leadership, commercial factors Player's potential (age) The length of time a player enters the NBA (the longer the time, the higher the salary limit) the player's injury player Race, nationality

2. Factors in the NBA league: the increase in TV broadcast fees, changes in other commercial factors, changes in the team’s salary cap (the team’s salary cap has been increasing in recent years, and the overall salary level of new players has been increasing) NBA style Changes (previously noticed inside, now pay attention to three points)

3. Other uncertain factors 　

2.1.3 System introduction: The NBA player salary prediction system mainly considers the impact of the player's technical statistics on their salary level, and predicts the player's salary based on the proportion of the salary cap (excluding the impact of the change of the salary cap). Get relevant datasets in Kaggle , Use Python to analyze the player's salary distribution; Analyze the various factors that affect the salary and the main influencing factors, provide a reference for the team managers and players themselves; Use a variety of machine learning algorithms to predict the salary, and choose the best method To provide a reference for establishing a salary system;

2.1.4 Data source: Kaggle 2019-2020 NBA player data, and data sets on websites such as ESPN.

2.2 Target Actors

Salary forecasting system customer: The customers of the salary prediction system mainly come from the management of the NBA team, such as the general manager of the team and related managers. They need a player’s future salary prediction and the value of the player’s contract extension, so all they have to do is enter the player’s name. After entering the system, the system will evaluate the player's previous game data, age, physical data and background information, and finally issue a complete salary forecast report to the customer.

Chatbot system: When the customer receives the player's salary prediction report, they may need some additional data to support such as processor data or the salary report of similar players. At this time, the system will have a chat robot to communicate with the customer and provide additional data. At the same time, the chat robot You can also transfer manual services to customers.