K-Means Clustering Practice¶

I used a dataset containing seed features and seed type (found here: https://data.world/databeats/seeds) to practice K-Means Clustering¶

Import Libraries¶

In [1]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt 
from mpl_toolkits.mplot3d import Axes3D
import seaborn as sns
from sklearn.cluster import KMeans 
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import accuracy_score
In [2]:
import warnings

warnings.filterwarnings('ignore')

Import and Inspect File¶

In [3]:
seeds = pd.read_csv('seeds.csv')
In [4]:
seeds.head()
Out[4]:
ID area perimeter compactness lengthOfKernel widthOfKernel asymmetryCoefficient lengthOfKernelGroove seedType
0 1 15.26 14.84 0.8710 5.763 3.312 2.221 5.220 1
1 2 14.88 14.57 0.8811 5.554 3.333 1.018 4.956 1
2 3 14.29 14.09 0.9050 5.291 3.337 2.699 4.825 1
3 4 13.84 13.94 0.8955 5.324 3.379 2.259 4.805 1
4 5 16.14 14.99 0.9034 5.658 3.562 1.355 5.175 1
In [5]:
pd.crosstab(index=seeds["seedType"],columns="count")
Out[5]:
col_0 count
seedType
1 70
2 70
3 70

The dataset is filled with 210 seed samples, seventy of each seed type.

Feature List¶

In [6]:
features = ['lengthOfKernel','asymmetryCoefficient','lengthOfKernelGroove']

I wanted to use only 3 features to make visualisation easy (obviously more could be used, and Principle Component Analysis could have determined which). Creating a features array allows me to change the features used in the model.

Scatter Plot of the Seeds Given The Feature List¶

In [7]:
fig = plt.figure(figsize=(10, 8))
ax = fig.add_subplot(111, projection='3d')

for seed_type, color in zip(seeds['seedType'].unique(), ['r', 'g', 'b']):
    x = seeds.loc[seeds['seedType'] == seed_type, features[0]]
    y = seeds.loc[seeds['seedType'] == seed_type, features[1]]
    z = seeds.loc[seeds['seedType'] == seed_type, features[2]]
    ax.scatter(x, y, z, c=color, label=f'Seed Type: {seed_type}')

ax.set_xlabel(features[0])
ax.set_ylabel(features[1])
ax.set_zlabel(features[2])
ax.legend()

plt.show()

Prepare the Data, and Run the K-Means Model¶

In [8]:
X = seeds[features]
y = seeds['seedType']
In [9]:
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)
In [10]:
kmeans = KMeans(n_clusters=3, random_state=42)
In [11]:
kmeans.fit(X_scaled)
Out[11]:
KMeans(n_clusters=3, random_state=42)
In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
KMeans(n_clusters=3, random_state=42)
In [12]:
cluster_centers = kmeans.cluster_centers_
print(cluster_centers)

[[-0.41346401 -0.78767024 -0.64725959]
 [ 1.21053806 -0.03869657  1.27322065]
 [-0.83969739  1.00570427 -0.62359456]]

Another Scatter Plot Allows Observation of the Centroids¶

In [13]:
fig = plt.figure(figsize=(10, 8))
ax = fig.add_subplot(111, projection='3d')

for seed_type, color in zip(y.unique(), ['r', 'g', 'b']):
    x = X_scaled[y == seed_type, 0]
    y_data = X_scaled[y == seed_type, 1]
    z = X_scaled[y == seed_type, 2]
    ax.scatter(x, y_data, z, c=color, label=f'Seed Type: {seed_type}')  # Updated label assignment

ax.set_xlabel(features[0])
ax.set_ylabel(features[1])
ax.set_zlabel(features[2])
ax.legend()

ax.scatter(cluster_centers[:, 0], cluster_centers[:, 1], cluster_centers[:, 2],
           c='black', marker='x', s=300, label='Cluster Centers')

# Show the plot
plt.show()

Determine the Accuracy of the Model (Comparing Predicted Seed Types to Actual Data)¶

Note: 1 has been added to the seed types below, as the the model produces (0,1 or 2) but the seed types are 1,2 or 3.

In [14]:
predicted_labels = kmeans.labels_
predicted_seed_types = predicted_labels + 1
In [15]:
print(predicted_seed_types)

[1 1 1 1 1 1 1 1 2 1 3 1 3 1 1 3 3 1 1 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
 1 1 3 1 1 1 2 1 1 1 1 1 1 1 3 1 1 1 1 1 1 1 1 1 1 1 3 1 1 1 1 1 1 2 2 2 2
 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
 2 2 2 2 2 2 2 2 2 2 2 2 2 1 2 2 2 2 2 2 2 2 2 2 1 2 2 2 2 3 3 3 3 3 3 1 3
 1 3 3 3 3 1 3 3 1 3 3 3 1 3 3 3 3 1 3 1 3 1 3 3 3 3 3 3 3 3 3 1 3 3 3 3 3
 3 3 3 3 3 3 3 1 3 3 3 3 3 1 1 3 1 3 3 3 1 3 3 1 3]
In [16]:
print(y)

0      1
1      1
2      1
3      1
4      1
      ..
205    3
206    3
207    3
208    3
209    3
Name: seedType, Length: 210, dtype: int64
In [17]:
accuracy = accuracy_score(y, predicted_seed_types)
In [18]:
print("Accuracy:", accuracy)

Accuracy: 0.8714285714285714

Making Prediction of A Future Seed¶

In [19]:
newSeedFeatures = [6,2,6] #['lengthOfKernel','asymmetryCoefficient','lengthOfKernelGroove']

newSeedFeaturesScaled = scaler.transform([newSeedFeatures])

predictedCluster = kmeans.predict(newSeedFeaturesScaled)

seedType = predictedCluster + 1

print("Predicted seed type:", seedType)

Predicted seed type: [2]

If We Hadn't Known There Were 3 Clusters, We Could Have Predicted This By Finding the Elbow On An Interia Plot¶

In [20]:
num_clusters = list(range(1, 9))
inertias = []
In [21]:
for k in num_clusters:
  model = KMeans(n_clusters=k)
  model.fit(X_scaled)
  inertias.append(model.inertia_)
In [22]:
plt.plot(num_clusters, inertias, '-o')

plt.xlabel('Number of Clusters (k)')
plt.ylabel('Inertia')

plt.show()