Merge pull request #642 from lck6055/knn-algo

Add KNN-Algo

Author: Karanjot786

Python/machine_learning/knn.py

@@ -0,0 +1,140 @@
+"""
+K-Nearest Neighbors (KNN) - Simple Implementation from Scratch
+-------------------------------------------------------------
+This script implements a basic version of the KNN algorithm for classification
+using only Python and NumPy (no sklearn).
+
+Concept Summary:
+----------------
+1. KNN is a supervised learning algorithm used for classification & regression.
+2. It finds the K nearest data points to a test point using a distance metric
+   (usually Euclidean distance).
+3. For classification → predicts the majority label among neighbors.
+4. For regression → predicts the average value among neighbors.
+5. It is a **lazy learner** (no explicit training phase, prediction happens at query time).
+
+Steps in this code:
+-------------------
+1. Compute Euclidean distance between the test point and all training points.
+2. Sort training points by their distance to the test point.
+3. Select top 'k' nearest points.
+4. Use majority voting to determine the predicted class.
+5. Return the predicted label.
+
+ Time Complexity (per prediction):
+        1 Distance calculation → O(n * d)
+        2 Sorting distances → O(n log n)
+        3 Selecting k nearest points → O(k)
+        4 Majority voting (Counter) → O(k)
+        🔹 Overall → O(n * d + n log n)
+
+    Space Complexity:
+        O(n * d) → to store all distances and intermediate computations
+
+Contributor:
+---------------------
+💻 Contributed by: **Lakhinana Chaturvedi Kashyap**
+"""
+
+import numpy as np
+import matplotlib.pyplot as plt  # ✅ Correct import
+from collections import Counter
+
+# Function: Euclidean Distance
+
+def euclidean_distance(p1, p2):
+    """
+    Calculates the Euclidean distance between two points.
+
+    Formula:
+    √( (x2 - x1)² + (y2 - y1)² + ... )
+    """
+    return np.sqrt(np.sum((np.array(p1) - np.array(p2)) ** 2))
+
+
+# Function: KNN Prediction
+
+def knn_prediction(training_data, training_labels, test_point, k):
+    """
+    Predicts the class of a test point using the K-Nearest Neighbors algorithm.
+    Returns both the predicted label and the k nearest points (for visualization).
+    """
+    distances = []
+    for i in range(len(training_data)):
+        dist = euclidean_distance(test_point, training_data[i])
+        distances.append((dist, training_labels[i], training_data[i]))  # include point itself
+
+    # Sort by distance (ascending)
+    distances.sort(key=lambda x: x[0])
+
+    # Select top k neighbors
+    k_neighbors = [label for _, label, _ in distances[:k]]
+    nearest_points = [point for _, _, point in distances[:k]]
+
+    # Majority voting
+    prediction = Counter(k_neighbors).most_common(1)[0][0]
+
+    return prediction, nearest_points
+
+# Example Usage
+
+# Convert to NumPy arrays for easy slicing
+training_data = np.array([
+    [1.0, 2.0],
+    [2.0, 3.0],
+    [3.0, 1.0],
+    [6.0, 5.0],
+    [7.0, 7.0],
+    [8.0, 6.0]
+])
+training_labels = np.array([0, 0, 0, 1, 1, 1])
+
+# Test data
+test_point = np.array([5.0, 5.0])
+k = 3
+
+# Predict
+prediction, nearest_points = knn_prediction(training_data, training_labels, test_point, k)
+
+print("Predicted label:", prediction)
+print("Nearest neighbors:", nearest_points)
+
+# Visualization
+
+plt.figure(figsize=(8, 6))
+
+# Plot class 0 points (blue)
+plt.scatter(
+    training_data[training_labels == 0][:, 0],
+    training_data[training_labels == 0][:, 1],
+    color='blue', label='Class 0', s=100
+)
+
+# Plot class 1 points (red)
+plt.scatter(
+    training_data[training_labels == 1][:, 0],
+    training_data[training_labels == 1][:, 1],
+    color='red', label='Class 1', s=100
+)
+
+# Highlight nearest neighbors (yellow)
+nearest_points = np.array(nearest_points)
+plt.scatter(
+    nearest_points[:, 0],
+    nearest_points[:, 1],
+    edgecolor='black', facecolor='yellow', s=200, label=f'{k} Nearest Neighbors'
+)
+
+# Plot test point (green star)
+plt.scatter(
+    test_point[0], test_point[1],
+    color='green', marker='*', s=250, label='Test Point (Predicted)'
+)
+
+# Labels and title
+plt.title(f"KNN Visualization (k={k}) — Predicted Label: {prediction}")
+plt.xlabel("Feature 1")
+plt.ylabel("Feature 2")
+plt.legend()
+plt.grid(True)
+plt.show()