Predicting NBA Player’s Points Scored

Slides:



Advertisements
Similar presentations
2008 FIBA RULES Presentation By: Marios Tenentes TABO Education Officer.
Advertisements

~KEVIN DURANT~ Greg Cannon. Kevin Durant is a great basketball player 1) The reason why he plays basketball 2.) Awards that he got in high school and.
Fantasy Football : NFL Score Predictor Matt Grecco Abhishek Goyal.
Lebron James vs. Carmelo Anthony Comparing the Two Rookie Seasons.
Basketball Audience: 6 th -12 th grade By: Lindsey Singer.
Using a Feed-forward ANN to predict NBA Games. About my ANN -Trained incrementally using back propagation -Currently it only uses sigmoid activation -Outputs.
Hitting One Out of the Park Presentation by: Richie Veihl Derek Monroe.
Basketball Stats Jeremy Lin is a point guard. And a point guard plays an important role on the basketball team. In some ways, the point guard is like.
 The objective of this report is to compare the statistics of five past and five current NBA players to find out who is the best of all time. PlayerGamesMPGFG%3P%FT%RPGAPGSPGBPGTOPPG.
Ball Speed: 2 kph Player Speed: 12 kph Closest Opponent: 7 m & behind Distance to Goal: 32m Chance of a Goal: Very High “Read the Game” by Sermetcan Baysal.
NBA Jeopardy Championships MVPS Coaches All Star Stats
NBA Statistics and Information Technology By: Greg Stitt.
Sports Communication Introduction Sports communication is very important in a game, especially in a basketball game. Without communication each team.
1 CSC 221: Introduction to Programming Fall 2013 Big data  building lists list comprehensions, throwaway comprehensions conditional comprehensions  processing.
3-1 6 th grade math Mean, Median, Mode and Range.
Forced and Unforced Errors Idea is originally from Tennis Major factor in Tennis.
Michael Muna.  Homeruns, RBI’s, Batting average  Run Created.
2014 World Cup Prediction Jared Ness. World Cup Play  Consists of 32 international teams  Must pass through 5 stages
Learning to Classify Documents Edwin Zhang Computer Systems Lab
Unit 4 Lesson 5 Demonstrating Mastery M.8.SP.1 To demonstrate mastery of the objectives in this lesson you must be able to:  Construct and interpret.
Unit 4 Lesson 6 Demonstrating Mastery M.8.SP.2 To demonstrate mastery of the objectives in this lesson you must be able to:  Know that straight lines.
Transfer Market Optimizer by Colton Freund and Zachary Krepps.
Using data to understand how you deliver software Or, a game where the players will try to predict a delivery
Greatest Player Of All Time? 3-Point Shooting Defense Scoring Free-Throw Shooting See For Yourself Winning By: Will Townsville.
BY ZACH ELSAWY. HOME RUNS HIT IN 2010 REGULAR SEASON.
USA BASKETBALL. NBA Most of the US players play in the NBA, because it is where the major players come from. The NBA( National Basketball Association)
Progress Report #2 By Adam Rothstein and Jesse Cox.
GAME TYPE STRATEGY Tailoring your strategy based on your game format.
Structures, Strategies and Compositions Gathering Data.
NBA Players’ Salaries Roy Philip Elal Index.
Jeremy Sylvain & Michael Schuckers
Chapter 7 Exploring Measures of Variability
NBA Players’ Salaries Stevan Dimitrijevic Start.
Conclusions and areas for further analysis
Statistically speaking… who were the top 5 players last season?
A Descriptive Model for NBA Player Ratings
Miami Heat Are you a fan of the #1 team in the NBA? Do you despise the Dallas Mavericks as much as I do for stealing the title away from the Miami Heat?
NBA: Analyzing Trends in NBA Officiating
QM222 Class 18 Omitted Variable Bias
Information Systems Development (ISD)
LEBRON JAMES BY: BEN ROBERTS.
1-Way Random Effects Model
Curling Statistics: How to Score
ANALYSING PERFORMANCE
NBA Draft Prediction BIT 5534 May 2nd 2018
Simple Linear Regression
Prediction of NBA games based on Machine Learning Methods
Machine Learning Week 1.
LeBron James By James Rogers.
Chapter 7 Part 1 Scatterplots, Association, and Correlation
Rodney J. Paul – Syracuse University
Dominic Hovasse and Jack Beach
Using Neural Networks to Determine NFL Game Outcomes
Winter-Quiz- Intermezzo
The Math of Baseball Will Cranford 11/1/2018.
BASKETBALL.
Dynamic Detection of Novice vs. Skilled Use Without a Task Model
Board Games! Jacob Simmons.
Machine Learning Application
CQ2 – What are the planning considerations for improving performance?
Chapter 3 Describing Relationships
Nba player By. Logan Ball..
Statistics Mean Absolute Deviation.
Basketball Position Classification
Mastering Open-face Chinese Poker by Self-play Reinforcement Learning
BASKETBALL.
Statistics Mean Absolute Deviation.
Loras College Men’s Basketball
Playing 25 Lines ( Total Bet 25 ).
Presentation transcript:

Predicting NBA Player’s Points Scored Doug Neu CS 539 UW Madison

Executive Summary Main Objective: Use a player’s previous seasons to predict how many points they will score in upcoming games. Useful for betting over/under on player points. Tasks Completed: Player stats scraped from NBA.com using their API Model created with top 12 players from previous season. Script written to iterate through all players and generate the projections Changes from Proposal: Decided k-means clustering wouldn’t be as useful as initially thought. Now a stretch goal. Decided on basing model off of only top tier players with consistent statistical output.

Data Collection & Platform Scraped player stats from NBA.com using their API. Scraped more specialized stats (Defensive rating, pace) from ESPN using Selenium. 2015-16, 2016-17 seasons stored as training data 2017-2018 stored as testing data Platform Laptop

Program Created a new script in Python Used Tensorflow and Keras. Linear Regression (still investigating Bayesian Ridge Regression)

Stats Used * Avoided using stats that are too result based Last 10 Avg FGA Last 10 Avg 3PA Last 10 Avg FTA Last 3 Avg Min Last 10 FG% Last 10 3P% Last 10 FT% Avg FGA for Season Avg FTA for Season Avg 3PA for season Avg Min for season Avg Pace of Game Opponent’s defensive rating * Avoided using stats that are too result based

Results MAE from Season AVG MAE from Individual Player Model (10 run AVG) MAE from All Player Model (10 run AVG) Anthony Davis 7.50 7.86 7.72 Blake Griffin 5.19 5.56 5.27 Damian Lillard 6.33 6.47 6.43 DeMar DeRozan 6.85 7.01 6.82 Giannis Antetokounmpo 5.57 6.52 5.81 James Harden 6.90 6.62 6.74 Kawhi Leonard 6.31 6.14 6.07 Kemba Walker 6.48 6.37 Kevin Durant 5.85 6.04 5.79 LeBron James 6.15 6.22 6.29 Zach Lavine 6.28 6.09

Error Report

Future Improvements Detect & eliminate extreme outliers as the result of Foul Trouble, Injuries, lack of minutes, coach’s decision. Detect if minutes are well below average  outlier Overtime can lead to a player playing 5 extra minutes  outlier Increased usage due to teammate injuries  outlier Add more players to model Investigate and add more features to model