Weekly schedule:

Module 1 (week 1-3) -- foundations

• Week 1 (Aug 22nd-)
  Topics:
  
  * Class overview & Intro to data mining (ppt)
  * Introduction to Kaggle and Python (notes)
  * Data (types of data, and data preprocessing) (ppt; pandas)
  * EDA & visualization (metaplotlib tutorials)
  
  Readings: Chapter 1 & 2
  More readings (Data Mining and health): Covid or flu?; genome/gene sequencing; ‘Major errors’ alleged in landmark study that used microbes to identify cancers
  Notebooks: Numpy & Pandas demo; ; Data exploration (EDA); Data preprocessing; EDA case study.

  ===================================

• Week 2 (Aug 29th-)
  Topics:
  
  * Where/how to get the data? (notes)
  * Vectors, matrices, and eigenvalue and eigenvector (notes)
  * Curse of dimensionality and dimensionality reduction (ppt)
  
  Readings: Tan Appendices A, B and C (pdf); (Massive) Chapter 11.
  Notebooks: SVD explained; PCA explained; SVD & PCA relationship; Standardization & PCA
  For bored: NMF; MDS; t-SNE (non-linear dimensionality reduction); the art of using t-SNE
  

  ===================================

• Week 3 (Sep 5th-)
  No class on Sep 5 (labor day)
  Topics:
  
  * Basics of probability and statistics (ppt)
  * Proximity measure (ppt)
  
  Reading: Chapter 2, (Massive) Chapter 3
  Notebooks: Data preprocessing (review); Standardization & PCA (review); Comparison of different scaling approaches
  

  ===================================

  Module 2 -- search & link analysis

• Week 4 (Sep 12th-)
  Topics:
  
  * KNN: Learning from neighbors (notes; KNN in sklearn; algorithms for near neighbor searches)
  * Finding similar items: Minhash & LSH (ppt)
  
  Reading: Chapter 5, (Massive) Chapter 3
  Notebooks: Fraud Detection using KNN; Work with texts

  ===================================

• Week 5 (Sep 19th-)
  Topics:
  
  * Finding similar items: Minhash & LSH (Cont.) (ppt; a practice)
  * Link analysis (PageRank): transition matrix and power iteration (ppt); Markov models (ppt)
  
  Reading: Chapter 6; Google SEO; google link schemes; qualify links; google spambrain
  Notebook: Facebook Network Analysis using NetworkX
  For bored: Twitter's "circle of trust" algorithm for WTF
  

  ===================================

  Module 3 -- association rule mining

• Week 6 (Sep 26th-)
  Topics:
  
  * Link analysis (PageRank): transition matrix and power iteration (ppt); Markov models (ppt)
  (Cont.)
  * Association analysis: Frequent itemset generation & rule generation (Apriori algorithm) (ppt) (an example)
  
  Reading: Chapter 7
  More readings (HPV vacination): paper abstract; statement of retraction; online post 1; post 2
  

  ===================================

• Week 7 (Oct 3rd-)
  Topics:
  
  * Association analysis: support counting (hash tree, and DHP); evaluation of association patterns (Simpson's paradox)
  * Association analysis: advanced concepts (non-binary data; concept hierarchy; sequential patterns; subgraph patterns) (ppt)
  * Genome-wise association studies (a paper)
  Video: explaining Simpson's paradox using covid-19
  Readings: DHP/SIT algorithms for association rules mining; Benefits and limitations of genome-wide association studies
  

  ===================================

  Module 4 -- clustering analysis

• Week 8 (Oct 10th-)
  Topics:
  
  * Cluster analysis (ppt)
  
  -- Clustering algorithms: k-means and variants, hierarchical clustering, density-based clustering
  -- Clustering evaluation: supervised vs unsupervised; heatmap, correlation, SSE, Silhouette coefficient, entropy, purity
  
  * Optional: Advanced clustering algorithms: EM algorithm, LSH & CURE (ppt)
  
  Reading: Chapter 8; clustering ensemble method; a paper on improving k-Means; Google News Personalization (MinHash clustering)
  Notebooks: some toy datasets; notebook (comparison of basic clustering algorithm) (scikit-learn); comparison of hierarchical clustering; silhouette analysis on sklearn
  Others: tree of coronavirus; clustering & data processing (recap)

  ===================================

  Module 5 -- classification (basic)

• Week 9 (Oct 17th-)
  Topics:
  
  * Classification: basic concept & techniques (decision tree; ppt); KNN (ppt)
  * Naive Bayes classifier (ppt)
  * Model selection/evaluation (ppt); overfitting/regularization (ppt)
  
  Notebook: Decision tree & pruning (breast cancer dataset 1; 2)
  Reading: Chapters 3 & 4
  For bored: XGBoost

  ===================================

  Midterm exam

• Week 10 (Oct 24th-)
  
  Review
  Rules

  Module 6 -- classification (cont.)

• Week 11 (Oct 31st-)
  Topics:
  
  * Model selection/evaluation (ppt); overfitting/regularization (ppt)
  * SVM (ppt)
  
  Notebook: Decision tree & pruning (breast cancer dataset 1; 2); SVM (GridSearchCV)
  Reading: Chapters 3 & 4
  For bored: XGBoost

  ===================================

• Week 12 (Nov 6th-)
  Topics:
  
  * Linear regression & logistic regression (ppt)
  * ANN (ppt-1; ppt-2); Deep Learning (e.g., CNN, ppt)
  * Ensemble Methods (Bagging & Boosting) (ppt)
  
  Reading: Chapter 4, 9; a bit of history
  Reading: pytorch vs tensorflow (google trends)
  Notebook: Linear regression; Fraud Detection (evaluation & logistic regression); Neural network models in scikit-learn
  Notebook (more): Voting classifier; A tutorial on Tensorflow: clothing classification using the Fashion MNIST dataset or text classification
  For bored: Opportunities and obstacles for deep learning in biology and medicine; Attention is all you need; Transformers at Huggingface; Facebook Research's FAISS
  

  ===================================

  Module 7 -- advertising on the web, recommendation system

• Week 13 (Nov 13th-)
  Topics:
  
  * Advertising on the Web (ppt; Alphabet Q3 FY23 income)
  * Recommendation systems (ppt)
  
  Reading 1: (Massive) Chapter 9; Surprise
  Reading 2:How google adwords auction work; How Data Mining Can Help Advertisers Hit Their Targets
  For fun: Touching the Void; Spotify active users; "Surprise me" recommendation

  ===================================

• Week 14 (Nov 20th-)
  Thanksgiving break; no class

  ===================================

  Module 8 -- everything else

• Week 15 (Nov 27th-)
  Topics:
  
  * Text Mining & NLP (notes)
  * Anormaly Detection (ppt)
  * Avoid False Discoveries (ppt)
  * Human factors, security and ethical issues in DM
  
  Notebooks: Anomaly Detection
  

  ===================================

• Week 16 (Dec 4th-)
  5-mins mini-presentations on projects (in class on both Tuesday and Thursday)