Back

2_Three_Main_Tasks

Loading views...
← PreviousNext →

Three Main Tasks in Machine Learning

Machine learning performs three primary tasks, each serving different purposes:

1. Prediction (Regression)

What is Prediction?

<abbr title="Predicting continuous numerical values like prices, temperatures, or quantities">Prediction</abbr> involves training a model on <abbr title="Data with known correct answers">labeled data</abbr> to forecast future continuous values.

How it Works:

  1. Collect historical data with known outcomes
  2. Train the model to find patterns and relationships
  3. Use the model to predict future values

Common Models:

  • <abbr title="Finding a straight line that best fits the data points">Linear Regression</abbr> - Predicts values using a straight line relationship
  • <abbr title="Finding a curved line that best fits the data points">Polynomial Regression</abbr> - Predicts values using curved relationships
  • Ridge Regression - Linear regression with penalty to prevent overfitting
  • Lasso Regression - Linear regression that can eliminate less important features

💻 Code Example: Linear Regression

import numpy as np
from sklearn.linear_model import LinearRegression

# Sample data (X: input, y: target)
X = np.array([[1], [2], [3], [4], [5]])
y = np.array([2, 4, 6, 8, 10])

# Create and train the model
model = LinearRegression()
model.fit(X, y)

# Make a prediction for input 6
prediction = model.predict(np.array([[6]]))
print(f"Prediction for input 6: {prediction[0]}") # Output: 12.0

Real-World Examples:

  • House Price Prediction - Predicting property prices based on size, location, age
  • Stock Market Forecasting - Predicting future stock prices
  • Weather Prediction - Forecasting temperature, rainfall
  • Sales Forecasting - Predicting future product sales
  • Energy Consumption - Predicting electricity usage

Key Characteristics:

  • Works with continuous numerical outputs (prices, temperatures, quantities)
  • Requires <abbr title="Data used to teach the model patterns">training data</abbr> with known values
  • Output is a number (not a category)

2. Classification

What is Classification?

<abbr title="Categorizing data into predefined groups or classes">Classification</abbr> involves training a model to categorize data into predefined groups or classes.

How it Works:

  1. Collect labeled examples from each category
  2. Train the model to recognize patterns for each class
  3. Use the model to classify new, unseen data

Common Models:

  • <abbr title="Tree-like model that makes decisions based on asking questions">Decision Trees (DT)</abbr> - Makes decisions by asking yes/no questions
  • <abbr title="Finds the best boundary line to separate different classes">Support Vector Machines (SVM)</abbr> - Finds optimal boundaries between classes
  • <abbr title="Computer systems inspired by the human brain with connected nodes">Artificial Neural Networks (ANN)</abbr> - Mimics human brain structure for complex patterns
  • Random Forest - Combines multiple decision trees for better accuracy
  • Naive Bayes - Uses probability to classify data
  • K-Nearest Neighbors (KNN) - Classifies based on similarity to nearby examples

Real-World Examples:

  • Email Spam Detection - Spam or Not Spam
  • Disease Diagnosis - Healthy or Diseased
  • Image Recognition - Cat, Dog, Bird, etc.
  • Sentiment Analysis - Positive, Negative, Neutral reviews
  • Credit Approval - Approve or Reject loan applications
  • Handwriting Recognition - Identifying written digits/letters

Key Characteristics:

  • Works with categorical outputs (labels, classes)
  • Requires labeled training data
  • Output is a category or class (not a number)

3. Clustering

What is Clustering?

<abbr title="Grouping similar data together without predefined categories">Clustering</abbr> involves grouping similar data points together without predefined categories, discovering natural patterns in <abbr title="Data without predefined categories or answers">unlabeled data</abbr>.

How it Works:

  1. Collect unlabeled data (no predefined categories)
  2. Apply clustering algorithm to find natural groupings
  3. Analyze clusters to understand patterns and similarities

Common Models:

  • K-Means - Divides data into K groups based on similarity
  • Hierarchical Clustering - Creates a tree of clusters
  • DBSCAN - Finds clusters of varying shapes and sizes
  • Gaussian Mixture Models - Uses probability distributions to cluster

💻 Code Example: K-Means Clustering

import numpy as np
from sklearn.cluster import KMeans

# Sample data (unlabeled)
X = np.array([[1, 2], [1, 4], [1, 0],
              [4, 2], [4, 4], [4, 0]])

# Create and fit the model (into 2 clusters)
kmeans = KMeans(n_clusters=2, random_state=0).fit(X)

# Predict the cluster for a new sample [0, 0]
cluster = kmeans.predict([[0, 0]])
print(f"Cluster for input [0, 0]: {cluster[0]}")

Real-World Examples:

  • Customer Segmentation - Grouping customers by shopping behavior
  • Document Organization - Grouping similar articles or documents
  • Image Segmentation - Separating different parts of an image
  • Anomaly Detection - Finding unusual patterns (fraud, defects)
  • Social Network Analysis - Finding communities in networks
  • Gene Sequence Analysis - Grouping similar genetic patterns

Key Characteristics:

  • Works with unlabeled data (no predefined answers)
  • Discovers hidden patterns automatically
  • Number of groups may or may not be known in advance

Comparison Table

Aspect Prediction Classification Clustering
Data Type Labeled Labeled Unlabeled
Output Continuous number Category/Class Groups/Clusters
Goal Forecast values Categorize data Find patterns
Example Predict price: $250,000 Spam: Yes/No Group 1, 2, 3
Learning Type Supervised Supervised Unsupervised
← PreviousNext →