Fall 2009 (Sep10-Dec7)
Faculty of Computer Science
Dalhousie University
|
|
[ http://www.cs.dal.ca/~vlado/hinf6210/coursecalendar.html ]
Fall 2009 (Sep10-Dec7) Faculty of Computer Science Dalhousie University |
| # | Date | Title | |
|---|---|---|---|
| Part I: Introduction | |||
| 1 | Th Sep 10 | Course Introduction
Course information: logistics and administrivia, textbook and main references, evaluation, policy on academic integrity, H1N1 influenza advisory, course content overview; Introductio to data mining: definition, descriptin of data mining, history, application examples: e-commerce, health informatics, and other application areas; data mining and other disciplines. Handouts: Course overview, tentative schedule, policy on academic integrity, H1N1 advisory, A0, slides Files: PDF slides. Reading: HK Ch.1 | A0 out |
| 2 | Tu Sep 15 | Data Mining Data Sources
DM as a step in KDD process, data mining and business intelligence, typical data mining system; different types of data: relational data, data warehouses, etc.; Data mining functionalities (or tasks): classification and prediction, data clustering Files: PDF slides. | |
| Part II: Fundamentals and Data Preprocessing | |||
| 3 | Th Sep 17 | Data Mining Functionalities
Data mining functionalities (continued): classification and prediction, cluster analysis, association rule mining, concept description: characterization and discrimination, outlier analysis, trend and evolution analysis, other; measures of interestingness; different views on data mining, major issues in data mining, major sources for research references; Course project: handout, deliverables; Part II: Fundamentals and Data Preprocessing, why is data preprocessing important, major tasks in data preprocessing, some statistical concepts, descriptive data summarization, measuring central tendency; distributive, algebraic, and holistic measures Files: PDF slides, Project Handout. Reading: HK Ch.2 | A0 due |
| 4 | Tu Sep 22 | Data Cleaning
Descriptive data summarization: measuring data dispersion, graphical summaries, major tasks in data preprocessing; data cleaning: missing values, noisy data, inconsistent data; simple binning: equi-width and equi-depth data; smooothing using binning methods; other types of smoothing; data integration: handling redundancy, correlational analysis, use of chi-square statistics in correlation analysis Files: PDF slides. | A1 out |
| 5 | Th Sep 24 | Data Transformation
Data transformation; normalization: min-max, z-score, decimal scaling, out-of-range values, softmax scaling, logistic function; Data reduction: strategies, compression: lossy and lossless compression, Discrete Wavelet Transformation, example; dimensionality reduction Files: PDF slides. | |
| 6 | Tu Sep 29 | Elements of Algorithmics
Brief algorithms review: algorithms, computational models: Turing machine and RAM, algorithm examples: mean, bubble sort, all subsets; algorithm complexity, polynomial vs. exponential algorithms, NP complete and NP hard problems; Methods for dimensionality reduction, feature selection (forward, backward, etc.), decision trees, principal component analysis, numerosity reduction, V-optimal and MaxDiff partitining, sampling Files: PDF slides. | |
| Tu Sep 29 | WEKA Tutorial (12:30-2:30pm, Teaching Lab 2, CSbldg) Files: PPT slides. | ||
| 7 | Th Oct 1 | Entropy-based Discretization
Sampling, sampling with and without replacement, stratified sampling; Discretization and concept hierarchies, types of attributes, discretization methods, segmentation by natural partitioning, some notions from information theory, entropy, expected value, discretization using entropy, examples; Files: PDF slides. | |
| Part III: Core Data Mining Methodology | |||
| 8 | Tu Oct 6 | Weka Overview
Concept hierarchies: specifications, automatic generation; Introduction to WEKA, some available tools in data mining, WEKA tool for data mining, WEKA: local resources, ARFF data format, running WEKA; Classification and Prediction; Chapter 6 overview, classification vs. prediction, supervised vs. unsupervised learning, classification process, evaluating classification models, classification with decision trees. handout: WEKA ARFF format and a WEKA tutorial Files: PDF slides. Reading: HK Ch.6 | A1 due |
| 9 | Th Oct 8 | Classification using Decision Trees
Classification with decision trees (cont.): example training data set, an algorithm for decision tree induction, ID3 algorithm, example, decision tree representation, avoiding overfitting in decision trees, other improvements; properties of decision tree algorithms; Bayesian Classification: basic notions in probability theory, examples Files: PDF slides. | P0 due |
| 10 | Tu Oct 13 | Naive Bayes Classification
Bayes theorem and its use in classification, Naive Bayes assumption and Naive Bayes classification, example, Naive Bayes classification with Gaussian distribution, handout, advantages and disadvantages of the Naive Bayes classification; Bayesian Networks, example Files: PDF slides. | A2 out |
| 11 | Th Oct 15 | Belief Bayesian Networks
Belief Bayesian Networks: definition, classification example (handout), learning Bayesian Networks; Classification by backpropagation: i.e., using artificial neural networks, background on artificial neural networks, an artificial neuron, multi-layer perceptron; Support Vector Machines (SVM): motivation, two-dimensional linearly separable case Files: PDF slides. | |
| 12 | Tu Oct 20 | Prediction
Support Vector Machines (cont'd): Multi-dimensional case, linearly inseparable case, introducing new columns, SVM vs. Neural networks, SVM related links; k Nearest Neighbour classification (kNN), Voronoi diagram; Other classification methods; Meta-classifiers (ensemble methods): bagging and boosting; Prediction: linear regression, 2 and more variables (multiple regression), non-linear regression; Distance measures, Minimization of different error distances Files: PDF slides. | |
| 13 | Th Oct 22 | Cluster Analysis
Logistic regression, predictor error measures, Using WEKA for classification, WEKA: ARFF data format, WEKA Classifiers, Command-line options, Command-Line options (2), examples of usage, examples of usage (2), Meta-learning schemes; Cluster Analysis (chapter 7): what is cluster analysis? general applications of clustering, examples of clustering applications, type of data in clustering analysis, distance measures Files: PDF slides. Reading: HK Ch. 7 | |
| 14 | Tu Oct 27 | Clustering Algorithms
Distance examples, cosine similarity and clustering of textual data; Major clustering approaches, partitioning algorithms; k-means clustring method: algorithm, example, strengths and weaknesses; K-Medoids method (PAM), Hierarchiclal clustering, density-based clustering: DBSCAN; Mining association rules (Chapter 5): support, confidence, item-sets, and frequent item-sets Files: PDF slides. Reading: HK Ch. 5 | A2 due |
| 15 | Th Oct 29 | Apriori Algorithm, Student presentations
Student presentations; Apriori algorithm: motivation and basic idea Files: PDF slides. | P1 due |
| Part IV: Data Warehouses and OLAP | |||
| 16 | Tu Nov 3 | Data Warehouses and OLAP
Apriori algorithm (continued): example, self-joining and pruning, challenges; Improvement of Apriori algorithm: partitioning with two database scans, FP-growth algorithm; Variations of association rule mining, multi-level and multi-dimensional rules, quantitative association rules, other variations; Concept description and comparison; Data Warehouses and OLAP: what is a data warehouse Files: PDF slides. Reading: [HK] Ch. 3, Ch. 4 | |
| 17 | Th Nov 5 | Conceptual Modelling of Data Warehouses
Data warehouse vs. operational DBMS; OLTP vs. OLAP; Multidimensional data model: lattice of cuboids, Conceptual modelling of data warehouses: star, snowflake, and fact fact constellation schema, examples, DMQL; Multidimensional data with concept hierarchies, OLAP operations Files: PDF slides. | A3 out |
| 18 | Tu Nov 10 | Efficient Cube Computation
OLAP operations (cont'd), Design of a data warehouse: a business analysis framework, process of data warehouse design, a three-tier warehouse architecture, OLAP server architectures, efficient data cube computation, cube operation; Multi-way array aggregation for cube computation, an efficient method for multi-way array aggregation for cube computation. Discussion about project presentations and project reports. Files: PDF slides. | A4 out |
| Part V: Course Review and Test | |||
| 19 | Th Nov 12 | Course Review
Course review, solutions to some sample problems, A2 solutions handed out Files: PDF slides. | |
| 20 | Tu Nov 17 | Test | |
| Part VI: Project Presentations | |||
| 21 | Th Nov 19 | Student presentations | A3 due |
| 22 | Tu Nov 24 | Student presentations | |
| 23 | Th Nov 26 | Student presentations | |
| 24 | Tu Dec 1 | Student presentations | |
| 25 | Th Dec 3 | Course Evaluation, Student presentations | |
| Mo Dec 7 | End of term, Project report due | Report due, A4 due | |