Software Automation
Scikit - KNN

Objective

By the end of this lab, you will be able to:

Use NumPy and scikit-learn to implement K-Nearest Neighbour classification
Visualise 2D data points and their classes
Make predictions using KNN and understand how k affects results

Part 1: Setup and Imports

Step 1: Install scikit-learn

Run this only once in your environment.

pip install scikit-learn

Step 2: Import Libraries

import numpy as np

import matplotlib.pyplot as plt

from sklearn.neighbors import KNeighborsClassifier

Part 2: Define and Visualise Data

We’ll classify points based on their location in 2D space.

Part 3: Train a KNN

# Create and train model with k=3

model = KNeighborsClassifier(n_neighbors=3)

model.fit(X, y)

Part 4: Making Predictions

# New point to classify

X_new = np.array([[5, 5]])

y_pred = model.predict(X_new)

print(f"Predicted class: {y_pred[0]}")

Part 5: Visualise the Prediction

plt.figure(figsize=(6, 6))

for i in range(len(X)):

colour = 'blue' if y[i] == 0 else 'red'

plt.scatter(X[i][0], X[i][1], color=colour, s=80)

# Plot new point

plt.scatter(X_new[0][0], X_new[0][1], color='green', s=100, marker='X', label='New Point')

plt.title("KNN Prediction")

plt.xlabel("Feature 1")

plt.ylabel("Feature 2")

plt.grid(True)

plt.legend()

plt.show()

Part 6 - Lab Exercises

Try Different Values of k

Change n_neighbors to 1, 3, 5, etc.

How does the prediction change?
Which value of k seems most stable?

Add New Training Points

Manually add more points to X and y. Try to:

- Add more class 0 and class 1 points in different areas
- Make the classes overlap and observe what KNN does
Use Real Categories

Create a KNN classifier for:

- “Is this fruit citrus?”
  1. Features: has thick skin, tastes sour, round shape (1 or 0)
  2. Labels: 1 = citrus, 0 = not citrus
- Invent 8 fruits with these properties and classify a new one

Page updated

Report abuse

Software AutomationScikit - KNN