Run in Google Colab

View on GitHub

Homework 8 - Clustering¶

In this assignment, you will be preforming the specified clustering methods in python

---

Step 1: Load Data¶

• Load the following attributes from Player_Attributes:
  • gk_reflexes
  • gk_kicking

These values will be used for clustering.

---

Step 2: Clustering¶

• Apply Kmeans (5 clusters), plot the results
• Use the Elbow technique, and report the optimal K value
• Run Kmeans again, with the k value from the previous step. Plot your results.

Dataset Overview¶

The dataset covers information about soccer players in sqlite format. This file is located in the Datasets directory of this repository. The file is called fifa_soccer_dataset.sqlite.gz. This is the same file from the previous homework (assignment 4).

If you haven't decompressed the file, you may need to follow the instructions below to decompress it.

IMPORTANT The database is compressed and needs to be decompressed before use. You can do this by running the following command in your terminal on Linux or MacOS:

gunzip Datasets/fifa_soccer_dataset.sqlite.gz

If you are using Windows, you can use the following command in your powershell:

$sourceFile = "$PWD\Datasets\fifa_soccer_dataset.sqlite.gz"
$destinationFile = "$PWD\Datasets\fifa_soccer_dataset.sqlite"

$inputStream = [System.IO.File]::OpenRead($sourceFile)
$outputStream = [System.IO.File]::Create($destinationFile)
$gzipStream = New-Object System.IO.Compression.GzipStream($inputStream, [System.IO.Compression.CompressionMode]::Decompress)
$gzipStream.CopyTo($outputStream)

$gzipStream.Close()
$outputStream.Close()
$inputStream.Close()

Alternatively, you can extract the file using the GUI of your operating system.

Submission Guidelines¶

• Submit your completed notebook as a HTML export, or a PDF file.

To export to HTML, if you are on Jupyter, select File > Export Notebook As > HTML.

If you are on VSCode, you can use the Jupyter: Export to HTML command.

• Open the command palette (Ctrl+Shift+P or Cmd+Shift+P on Mac).
  • Search for Jupyter: Export to HTML.
  • Save the HTML file to your computer and submit it via Canvas.

---

In [1]:

import sqlite3
import pandas as pd
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt

import sqlite3 import pandas as pd from sklearn.cluster import KMeans import matplotlib.pyplot as plt

In [33]:

dataset_path = "../../Datasets/fifa_soccer_dataset.sqlite" # Fix your path accordingly

# Your Code Here

dataset_path = "../../Datasets/fifa_soccer_dataset.sqlite" # Fix your path accordingly # Your Code Here

First, let's connect to the database as we've done in previous guides below.

In [76]:

conn = sqlite3.connect(dataset_path)

conn = sqlite3.connect(dataset_path)

From the database, grab gk_reflexes and gk_kicking from the Player_Attributes table.

In [ ]:

# get only `gk_reflexes` and `gk_kicking
player_attr_df = ...

# get only `gk_reflexes` and `gk_kicking player_attr_df = ...

Now, let's drop the missing values.

In [93]:

# drop the null values
player_attr_df = player_attr_df.dropna()
player_attr_df.head()

# drop the null values player_attr_df = player_attr_df.dropna() player_attr_df.head()

Out[93]:

	gk_reflexes	gk_kicking
0	8.0	10.0
1	8.0	10.0
2	8.0	10.0
3	7.0	9.0
4	7.0	9.0

Let's run the KMeans() function, using n_clusers = 5 as our starting cluster count below.

In [94]:

km = KMeans(n_clusters = 5, init = 'random', n_init = 10, max_iter= 300, tol=1e-04, random_state=1)
ykm = km.fit_predict(player_attr_df)

km = KMeans(n_clusters = 5, init = 'random', n_init = 10, max_iter= 300, tol=1e-04, random_state=1) ykm = km.fit_predict(player_attr_df)

Plot the results using the scatter() function, passing player_attr_df.gk_reflexes and player_attr_df.gk_kicking as the first two arguments.

In [ ]:

plt.scatter(_, _, c=ykm, marker='o', s = 50, cmap='tab10')
centers = km.cluster_centers_
plt.scatter(centers[:,0], centers[:,1], c ='black', s = 200)

plt.grid()
plt.tight_layout()
plt.show()

plt.scatter(_, _, c=ykm, marker='o', s = 50, cmap='tab10') centers = km.cluster_centers_ plt.scatter(centers[:,0], centers[:,1], c ='black', s = 200) plt.grid() plt.tight_layout() plt.show()

Above we can see the 5 centroids from the clustering. It looks like we can probably choose a better cluster number. Let's run the elbow method below to find a better cluster count. Once done, plot the distance in the plt.plot() so we can see which value to use.

In [ ]:

dist= []
for i in range(1, 11):
    km = KMeans(n_clusters = i, init = 'k-means++', n_init = 10, max_iter= 300, random_state=0)
    km.fit(player_attr_df)
    dist.append(km.inertia_)

# Pass dist where the parameter underscore is below
plt._(range(1,11), _, marker = 'o')
plt.tight_layout()
plt.show()

dist= [] for i in range(1, 11): km = KMeans(n_clusters = i, init = 'k-means++', n_init = 10, max_iter= 300, random_state=0) km.fit(player_attr_df) dist.append(km.inertia_) # Pass dist where the parameter underscore is below plt._(range(1,11), _, marker = 'o') plt.tight_layout() plt.show()

Looks like we should use 2! In the cell below, run KMeans() again, same as previously in this guide, and use 2 clusters.

In [ ]:

km = # Run KMeans again here
ykm = # Fit Predict Here

km = # Run KMeans again here ykm = # Fit Predict Here

Lastly, plot your results.

In [1]:

# Plot as we did before

# Plot as we did before