Project: ML - K-means clustering (Senator Votes and NBA players)


Problem 1:

  • Find clusters in Senator Votes from data on yes and no on votes for each senator

Problem 2:

  • Find clusters in NBA players from data on : assist/turnover ratio and points per game

Tools:

  • Clustering with K-Means, find distances to cluster centres, visualize clusters


load defaults

In [1]:
import numpy as np
import pandas as pd
import seaborn as sns
import re
import requests 
import math

%matplotlib inline
import matplotlib.pyplot as plt
from matplotlib.ticker import MultipleLocator
from matplotlib import rcParams
import matplotlib.dates as mdates
from datetime import datetime
from IPython.display import display, Math

from functions import *

plt.style.use('seaborn')
plt.rcParams.update({'axes.titlepad': 20, 'font.size': 12, 'axes.titlesize':20})

colors = [(0/255,107/255,164/255), (255/255, 128/255, 14/255), 'red', 'green', '#9E80BA', '#8EDB8E', '#58517A']
Ncolors = 10
color_map = plt.cm.Blues_r(np.linspace(0.2, 0.5, Ncolors))
#color_map = plt.cm.tab20c_r(np.linspace(0.2, 0.5, Ncolors))


#specific to this project
from sklearn.cluster import KMeans

print("Defaults Loaded")

Defaults Loaded


K-means clustering, 2 clusters


Dataset: congress votes

In [2]:
votes = pd.read_csv('./data/114_congress.csv')
display(votes[:3])
name party state 00001 00004 00005 00006 00007 00008 00009 00010 00020 00026 00032 00038 00039 00044 00047
0 Alexander R TN 0.0 1.0 1.0 1.0 1.0 0.0 0.0 1.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0
1 Ayotte R NH 0.0 1.0 1.0 1.0 1.0 0.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0
2 Baldwin D WI 1.0 0.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0 0.0 1.0 1.0 0.0 1.0 1.0
In [3]:
print(votes['party'].value_counts())

R    54
D    44
I     2
Name: party, dtype: int64
In [4]:
#fit transform returns in the eucledian distance from each Senator to the first cluster and second clusters 

kmeans_model = KMeans(n_clusters=2, random_state=1)
senator_distances = kmeans_model.fit_transform(votes.iloc[:,3:])

#get clustering labels from model
labels = kmeans_model.labels_
#compare model with original data
display(pd.crosstab(labels,votes['party']))
party D I R
row_0
0 41 2 0
1 3 0 54

some democrats voted along side the Republicans in some votes, identify the outliers:

In [5]:
democratic_outliers = votes[(labels==1) & (votes['party']=='D')]
sel = votes['party']=='D'
print("Average vote from Democrats in first votes:")
print(votes[sel].iloc[:,:9].mean())
print(democratic_outliers.iloc[:,:9])

Average vote from Democrats in first votes:
00001    0.704545
00004    0.079545
00005    0.011364
00006    0.988636
00007    0.011364
00008    0.875000
dtype: float64
        name party state  00001  00004  00005  00006  00007  00008
42  Heitkamp     D    ND    0.0    1.0    0.0    1.0    0.0    0.0
56   Manchin     D    WV    0.0    1.0    0.0    1.0    0.0    0.0
74      Reid     D    NV    0.5    0.5    0.5    0.5    0.5    0.5
In [6]:
plt.scatter(senator_distances[:,0],senator_distances[:,1],c=labels)
plt.show()

find rows further away from other cluster:

In [7]:
extremism = np.sum(senator_distances**3, axis=1)
votes['extremism'] = extremism
votes.sort_values('extremism', ascending = False, inplace=True)
print(votes.iloc[:10,[0,1,2,3,4,5,6,7,8,-1]])

         name party state  00001  00004  00005  00006  00007  00008  extremism
98     Wicker     R    MS    0.0    1.0    1.0    1.0    1.0    0.0  46.250476
53   Lankford     R    OK    0.0    1.0    1.0    0.0    1.0    0.0  46.046873
69       Paul     R    KY    0.0    1.0    1.0    0.0    1.0    0.0  46.046873
80      Sasse     R    NE    0.0    1.0    1.0    0.0    1.0    0.0  46.046873
26       Cruz     R    TX    0.0    1.0    1.0    0.0    1.0    0.0  46.046873
48    Johnson     R    WI    0.0    1.0    1.0    1.0    1.0    0.0  40.017540
47    Isakson     R    GA    0.0    1.0    1.0    1.0    1.0    0.0  40.017540
65  Murkowski     R    AK    0.0    1.0    1.0    1.0    1.0    0.0  40.017540
64      Moran     R    KS    0.0    1.0    1.0    1.0    1.0    0.0  40.017540
30       Enzi     R    WY    0.0    1.0    1.0    1.0    1.0    0.0  40.017540


K-means Clustering


Dataset: NBA stats on players

In [8]:
nba = pd.read_csv("./data/nba_2013.csv")
display(nba.iloc[:3,:20])
player pos age bref_team_id g gs mp fg fga fg. x3p x3pa x3p. x2p x2pa x2p. efg. ft fta ft.
0 Quincy Acy SF 23 TOT 63 0 847 66 141 0.468 4 15 0.266667 62 126 0.492063 0.482 35 53 0.660
1 Steven Adams C 20 OKC 81 20 1197 93 185 0.503 0 0 NaN 93 185 0.502703 0.503 79 136 0.581
2 Jeff Adrien PF 27 TOT 53 12 961 143 275 0.520 0 0 NaN 143 275 0.520000 0.520 76 119 0.639

point guards are crucial because they create scoring opportunities, lets look at what types of point guards exist and group similar point guards together

  • look at points per game
  • $ATR = \frac{Assists}{Turnovers}$
In [9]:
point_guards = nba[nba['pos']=='PG'].copy()

#ppg
point_guards['ppg'] = point_guards['pts'] / point_guards['g']

#drop players with no turnovers
point_guards = point_guards.loc[point_guards['tov'] != 0]
point_guards['atr'] = point_guards['ast']/point_guards['tov']

plt.scatter(point_guards['ppg'], point_guards['atr'], c='y')
plt.title("Point Guards")
plt.xlabel('Points Per Game', fontsize=13)
plt.ylabel('Assist Turnover Ratio', fontsize=13)

plt.show()

maybe 5 clusters (definitely 3)

In [14]:
num_clusters = 5
In [15]:
# Visualizing clusters
def visualize_clusters(df, num_clusters):
    colors = ['b', 'g', 'r', 'c', 'm', 'y', 'k']

    for n in range(num_clusters):
        clustered_df = df[df['cluster'] == n]
        plt.scatter(clustered_df['ppg'], clustered_df['atr'], c=colors[n-1])
        plt.xlabel('Points Per Game', fontsize=13)
        plt.ylabel('Assist Turnover Ratio', fontsize=13)
    plt.show()
In [16]:
kmeans = KMeans(n_clusters=num_clusters)
kmeans.fit(point_guards[['ppg', 'atr']])
point_guards['cluster'] = kmeans.labels_

visualize_clusters(point_guards, num_clusters)
In [ ]: