Big-Data Analytics for Cloud, IoT and Cognitive Computing

5Unsupervised Machine Learning Algorithms

CHAPTER OUTLINE

1. 5.1 Introduction and Association Analysis
   1. 5.1.1 Introduction to Unsupervised Machine Learning
   2. 5.1.2 Association Analysis and A priori Principle
   3. 5.1.3 Association Rule Generation
2. 5.2 Clustering Methods without Labels
   1. 5.2.1 Cluster Analysis for Prediction and Forecasting
   2. 5.2.2 K-means Clustering for Classification
   3. 5.2.3 Agglomerative Hierarchical Clustering
   4. 5.2.4 Density-based Clustering
3. 5.3 Dimensionality Reduction and Other Algorithms
   1. 5.3.1 Dimensionality Reduction Methods
   2. 5.3.2 Principal Component Analysis (PCA)
   3. 5.3.3 Semi-Supervised Machine Learning Methods
4. 5.4 How to Choose Machine Learning Algorithms?
   1. 5.4.1 Performance Metrics and Model Fitting
   2. 5.4.2 Methods to Reduce Model Over-Fitting
   3. 5.4.3 Methods to Avoid Model Under-Fitting
   4. 5.4.4 Effects of Using Different Loss Functions
5. 5.5 Conclusions

5.1 Introduction and Association Analysis

In this chapter, we introduce several major classes of unsupervised machine learning algorithms. First, we study data association analysis methods for unsupervised machine learning. Other unsupervised and semi-supervised machine learning methods are presented in subsequent sections. Machine learning performance metrics and a selection method for various ML algorithms are given in Section 5.4.

5.1.1 Introduction to Unsupervised Machine Learning

In unsupervised learning, the learner must work out a function to describe a hidden structure from unlabeled data. Since the ...