Specy
.app
Specy

Data Mining

Lecture notes covering the core techniques of data mining: data preparation and cleaning, similarity and distance functions, association rule mining, clustering algorithms, and anomaly detection methods.

Data Mining

Notes from the Data Mining course, one note per lecture. Each file integrates the lecture content with explanations and diagrams.

The course covers the full data mining pipeline: from understanding different data types and their dependencies, through the preparation and cleaning stage, to the four core analytical problems — association pattern mining, clustering, classification, and outlier/anomaly detection. Scalability challenges and streaming scenarios are discussed throughout.

Contents

  1. Introduction to Data Mining Overview of data mining as a discipline: categories of problems (association, clustering, classification, outlier detection), data types, dependency-oriented data, big data fundamentals, and scalability challenges.
  2. Data Preparation Covers the multistage data preparation pipeline: feature extraction, data type portability, cleaning (missing values, normalization), and data reduction techniques including sampling, feature selection, and dimensionality reduction.
  3. Similarity and Distances Defines and compares distance and similarity functions for quantitative, categorical, text, binary, time-series, and graph data, including Lp-norms, the curse of dimensionality, cosine similarity, Jaccard coefficient, DTW, and graph kernels.
  4. Association Mining Covers frequent itemset mining and association rule generation, including support, confidence, and interest metrics, along with efficient algorithms: A-Priori, PCY, Multi-Stage, Multi-Hash, and SON for distributed datasets.
  5. Clustering Analysis Explores clustering types (partitional vs. hierarchical), cluster definitions (well-separated, center-based, density-based), and key algorithms including agglomerative hierarchical clustering, K-Means, and DBSCAN.
  6. Anomaly Detection Covers detection of outliers in data: types of anomalies (point, contextual, collective), supervised/semi-supervised/unsupervised approaches, classification-based techniques, SVM, KNN, and the Local Outlier Factor (LOF) algorithm.