DSC 478 Programming Data Mining Applications Course Summary

DSC 478 Programming Data Mining Applications Course Summary Bamshad Mobasher De. Paul University

What we did i Data Preprocessing and Understanding 4 Data selection/preprocessing/transformation h Using Python, Numpy, Pandas, Scikit-learn 4 Emphasis on visualizing and understanding characteristics of the data h Primarily using Matplotlib i Supervised Knowledge Discovery 4 Classification h KNN, decision trees, naïve Bayes, LDA, Support Vector Machines, etc. 4 Regression Analysis h Linear regression, regularized regression, regression via gradient descent, KNN 4 Lots of emphasis on model evaluation h Evaluation metrics h Train-Test methodologies such as cross-validation h Systematic parameter selection (e. g. , grid search) h Feature selection 4 Ensemble methods (e. g. Random Forest, Ada. Boost) 2

What we did i Unsupervised Knowledge Discovery 4 Cluster analysis 4 Using PCA and SVD for dimensionality reduction, data characterization, and noise reduction 4 Emphasis on using unsupervised approaches as components of larger knowledge discovery efforts h E. g. , using PCA before clustering; using clustering as the basis for classification i Real application domains 4 Text Mining and document analysis/filtering h Content analysis, TFx. IDF transformation, text categorization, document clustering 4 Recommender systems h Neighborhood methods (user- and item-based) h Matrix factorization 4 Image analysis h Image classification and image clustering 4 Marketing 3

What we did not do (and you should learn later) i Approaches for mining sequential/temporal data 4 Markov models; time series analysis, sequential pattern mining i More Ensemble and Hybrid Classifiers/Predictors 4 Combining multiple classifiers i Other Kernel-Based Methods 4 Kernelized Regression 4 Kernel PCA i Topic modeling with Latent factor models 4 Latent Dirichlet Allocation i Neural Network Models and Deep Learning 4 Auto encoders 4 Convolutional NNs 4 Recurrent Neural Networks 4 …. 4