Source Of Summary:

Handout 7 - Feature Selection and Reduction.pdf

Handout 8 - reduced Feature Engineering.pdf

Feature Selection & Reduction Techniques & Applications 🚀

1. Introduction & Why Dimensionality Reduction? 💡

• Why Dimensionality Reduction?
  • Visualization: Projection of high-dimensional data onto 2D or 3D ✅
  • Data Compression: Efficient storage and retrieval ✅
  • Noise Removal: Positive effect on query accuracy ✅

2. Application of Dimensionality Reduction 📚

• Applications:
  • Customer Relationship Management: Enhance analysis and prediction ✅
  • Text Mining: Improve document analysis and classification ✅
  • Image Retrieval: Accelerate image searches ✅
  • Microarray Data Analysis: Enable better bioinformatics insight ✅
  • Protein Classification: Assist in the categorization of protein types ✅
  • Face Recognition: Allow robust recognition systems ✅
  • Handwritten Digit Recognition: Improve pattern recognition systems ✅
  • Intrusion Detection: Enhance security by detecting anomalies ✅

3. Document Classification Example 🔎

• Below is a snapshot of an example document classification scenario:

4. Gene Expression Microarray Analysis 🧬

• Overview:
  • Task: To classify novel samples into known disease types (disease diagnosis) ✅
  • Challenge: Thousands of genes, few samples ✅
  • Solution: Apply dimensionality reduction ✅
• Data Example – Expression Microarray Data Set:
  • Gene IDs: M23197_at, U66497_at, M92287_at
  • Samples:
    • Sample 1: 261, 88, 4778, … (Class: ALL)
    • Sample 2: 101, 74, 2700, … (Class: ALL)
    • Sample 3: 1450, 34, 498, … (Class: ALL)
• Additional Note:
  • Expression Microarray image (1.28cm, courtesy of Affymetrix)
  • AML and other types of high-dimensional data (face images, handwritten digits, etc.) ✅

5. Feature Reduction vs. Feature Selection 🔍

• Feature Selection:
  • Only a subset of the original features are selected ✅