Feng's Final Project and Reflection

Code and introduction

Here is my code

I plan to work on data analysis project on NBA players data from 2021-2022 season. Analyze player’s efficiency and provide advanced stats through data visualization library (matplotlib).

The reason why I am doing this project is that I am such a big basketball fan, I would like to use this opportunity to gain more insights from the players’ stats and help others to interpret the raw data more easily.

Milestones

I will illustrate my approach by describing how I completed evert milestone.

Find valid datasets from either Kaggle.com or other 3rd party gathered datasets.
1. Found 2 datasets from Kaggle.com called nba2021.csv and nba2k20-full.csv.
2. nba2021.csv provides basic data for each players played for NBA in 2020-21 season.
3. nba2k20-full.csv provides data such as salary, height, draft order etc for each player in 2020-21 season too.

[x]Start the data cleaning/exploratory.

Here is how I cleaned and merge two datasets into a single dataframe in preprocess.py:

def merge_n_clean(x,y,s):
  # select needed columns and merge these two dataframe by player's name
  filter = ['Player', 'Pos', 'PTS', 'TRB', 'AST', 'STL', 'BLK', 'FGA', 'FG', 'FTA',
       'FT', 'TOV']
  df = pd.merge(
    x[filter],
    y[["Player","Salary"]],
    on = 'Player'
  )
  # remove the dollar sign from salary column then turn it into interger
  df.Salary = [tmp.strip('$') for tmp in df.Salary]
  df['Salary'] = df['Salary'].astype(int)
  df = df.sort_values(by="Salary", ascending = False)
  return df

[x]Think of what kind of advanced statistic results I would like to visualize and can provide meaningful insights.

Since NBA player get paid so much, I think to do an evaluation between players salary and their EFF value on the court throughout the season would be fun. EFF value is not provided from any of the datasets that I found, so I have to calculate it myself using Python. That’s see the definition of EFF value first : EFF

Basically it is -> (PTS + REB + AST + STL + BLK − ((FGA − FGM) + (FTA − FTM) + TOV))

Since these stats can be found in nba2021.csv, I can easily calculate it using the following function and sorted the dataframe by EFF in calc.py:

  
def calc_EFF(df):
  
  df['EFF'] = (df['PTS'] + df['TRB'] + df['AST'] + df['STL'] + df['BLK']
  - (df['FGA'] - df['FG'] ) 
  + (df['FTA'] - df['FT'] ) 
  +df['TOV']
  )
  df = df.sort_values(by="EFF", ascending = False)
  
return df

After that, I would be able to caculate the ratio between salary and EFF for each player in calc.py :

  
# caculate cost effective ratio -> salary / EFF and sorted the dataframe by Ratio

def calc_ratio(df , order):
  
  df['Ratio'] = df['EFF'] / (df['Salary']/ 10000)
  if order == "ascending":
    df = df.sort_values(by="Ratio", ascending = True)[:10]
  else:
    df = df.sort_values(by="Ratio", ascending = False)[:10]
  return df  

Design and develop interface I used infinite while loop to create a control panel which can be broke by selectiing option 5

options ={
  "0" : "Print all Stats",
  "1" : "Top 10 Highest Salary Players",
  "2" : "Top 10 Highest EFF Players",
  "3" : "Top 10 Highest Cost-Effective Players",
  "4" : "Top 10 LEAST Cost-Effective Players",
  "5" : "Exit",
}

Each option provides different data query and charts.

Visualize the data I used matplotlib to visualize my data in bar charts as follows:
```
#Function to create a bar chart                
  def bar_plot(x, y, title, df):
    df.plot(kind = 'bar',
          x = x, # data for x axis
          y = y, # data for y axis
          title=title
          )
    # show the plot
plt.show()
```
And this is one of example in my code to do the visualization :
```
  bar_plot("Player","Ratio","Top 10 LEAST Cost-Effective Players",ratio)
```
Reflection

It’s very cool to get to utilize what we’ve learned from class to build up this project. While dealing with the project, I accidentally closed the tab for trinket when I almost finished it. Which is extremely frustrating , but I took it as a great opportunity to gain more hands on experience by code it from scratch once again. I was initially more familiar with plotly instead of matplotlib, however, trinket only supports matplotlib which forced me to use it. It’s also something valuable to me since if it were not for the project, I wouldn’t have the chance to get to know the library and gain familiarity with it.

Working on a project that suits my interest in always fun, I am glad that I get to do a project for class using the data from NBA and create meaningful insight.

If I had more time, I would want to try and create a search function that enable user to search for specific stats or players. I would want to provide more advanced stats and chart as well. Maybe doing something for major leagur sports such as MLB, NFL or even college football would be fun too.

Thank you Professor and all the classmates that have helped me throughout the way. I have been enjoying this class and learned a lot.