Project Overview
Project lifecycle
- Business Understanding
We are trying to access a shooters (overall) performance. Look at the same position and group players that are in the same category!
Chris Mullen 51% 2 point field goal, 38.4% 3 point field goal, 87% free throw Hypothentical (88 Season Chris Mullen 54% 2 point, 43% three point, 89% free throw) Hypothetical (08 season Malcom Brogdon 51% 2 point, 40% three point, 90% free throw)
I want to show that Chris Mullen had a better year because his performance was substatially better than Steph Curry in two categories.
50 - 40 - 90 (8 players have made this happen and 5 of these players were league MVP’s) only shows one season Steve Nash Kevin Durant Larry Bird Mark Price Steph Curry Malcom Brogdon Dirk Nvisky Reggie Miller
Chris Mullen What is efficiency?
- Performance
- Trying to define efficiency
Chris Mullen is a different type of player than Steph Curry. Chris Mullen is more of a set shooter, 1 dribble shooter, or 2 dribble shooter Steph Curry is a shooter that will fade away, shoot runners, floaters, off balanced shots, etc.
Chris Mullen is also 6’8 Steph Curry is 6’3
Chris Mullen is different than a player like Allen Iverson as well. He is taller and gets his shots in different ways. Allen IVerson is a volume shooter and was not that efficient.
I want to assume that these players best season shooting was their 50-40-90 year.
Another player may have shot substatially better then these players by a 1% point or more for a duration of a season.
The assumption is Steph Curry is the greatest shooter of all time and I am trying to prove that wrong by looking at his best season (50-40-90) and finding someone that shot it better than he did in their best season. I also have to determine what makes a player a great shooter. The people who I select will be a similar player and position so the model is scalable.
Right now I am thinking about Steve Nash and Mark Price to start out.
- Data Mining
Do we have the data of what type of shot it is? Where can we find this? Here is advanced stats from 1950-2017. NBA Stats - Kaggle
- Data Cleaning
I will be looking into data cleaning this weekend
- Data Exploration
Pull up as much data as I can find and find out what each player We want season data for those players
- Feature Engineering
Read blog post
- Predictive Modeling
- Data Visualization
I am looking to talk to a NBA data scientist about this project.
This is a good study on field goal percentage!
Thinking through machine intelligence
Taste
- Not relivent
Hear
- Shooting - We can use the sound of the net
- Shooting - Crowd noise
- The rim, the backboard, the net or the ground, someone catching a shot, makes a sound when you make or miss a shot.
- Referees whistle
- Buzzer
- Coaches, Players, etc talking
- Announcer
- Shot clock violation
- shoes squeeking
Touch
- You need the ball to go through the hoop for the shot to be successful
- Need to hold the ball
- You need to shoot the ball
- You need to put rotation on the ball
- YOu need the ball to hit the hoop net or backboard for it to go in
Smell
- Not relivent
See
- Being able to see when a person is shooting in the game is important to us
- Calculating shooting percentages
- By being able to visualize what is happening shooting wise on the court for an individual player
- Disingish what is going on on the court
- I know what is going on on the court but the computer needs to know what is going on on the court (how to distingish in our video all the other shooting from everything else. This is important so that we can extract the shooting data.
- Michael Jordan sticks out his tongue he is probably going to take a shot
Other things to think about:
- Sometimes the person who goes up to shoot ends up being the passer
- Sometimes a passer can make the shot on accident
- If a person accidently makes a shot should it negatively effect them or not
Who is the best point guard shooter of all time?
How can I create a model useful for my business?
We will start with these five players.
Steph Curry, Damion Lillard, Mark Price, Steve Nash, Jason Terry
Looking at data and try to figure out how I can use it. Make sure I focus on the shot plot and finding how to get the x,y axis. We have to find this information. EDA and visualizations to try to get to a point where I can graph the data.
I am looking to start with the x(1d), move to the x,y(2d) and then move the the x,y,z(3d) in that order - It gets progressively difficult in that order. This is what a data scientist named Rapheal told me to do.