CS5312 - Big Data Analytics

---

Course Description:
This course introduces the essential mathematical concepts and computational methods to store, retrieve, and process to store, retrieve, process, and analyze data sets that are too large, complex, and heterogeneous for conventional methods. We will learn how to handle different types of big data, such as text, streaming, or network data. The course covers data preprocessing, data visualization, text analytics, graph analytics, recommendation systems, clustering, classification, regression, dimensionality reduction, matrix factorization, and locality-sensitive hashing.

Prerequisites:

• An undergraduate course in Discrete Mathematics, Data Structures, Algorithms and Probability.
• An undergraduate course in Databases and Linear Algebra is useful but not required.
• Strong problem solving skills and some background in programming.

Textbook(s)/Supplementary Readings:

• Mining of Massive Datasets Jure Leskovec, Anand Rajaraman, Jeffrey D. Ullman

Lectures:

Topic	Slides	Notes
Introduction to Big Data Analytics Big Data Generation and Growth Industries benefiting from Data Analytics Sources, Aspects and Types of Big Data The Analytics Process	Slides
Getting to Know Data and EDA EDA: Purpose & Benefits Statistical EDA Graphical EDA	Slides
Data Preprocessing and Transformation Issues with Data Data Cleaning: Missing value, Noise, Outliers Data Integration: Inconsistencies, Deduplication Data Reduction: Sampling, Feature Selection Data Transformation: Standardization, Numeric transformation	Slides
Vector Norms and Proximity Measures Vector Norms and Unit Circles Proximity Measures Distance Between Non-numeric Vectors Distance Between Mixed Feature Vectors Distance Between Non-vectors Data Objects	Slides
Data Analytics Tasks and Methods Classification: Motivation and Applications Train-Validation Split & Cross-Validation Evaluation Metrics and Class Imbalance Overfitting Classifiers: k-NN, Naïve Bayes, Decision Tree Information Gain & Entropy Regression: Linear Regression Fitting Zero-degree Function and Line Interpreting Coefficients Partitioning the Variance Multiple Linear Regression Polynomial Regression Logistic Regression Clustering: Aspects and Definition Point Assignment Methods: k-Means, k-Medoid and k-Mode Agglomerative Clustering Clustering Validation and Evaluation External and External Measures Recommendation Systems: Raw Averages based Recommendation ANOVA and Bayesian Filtering Content Based Filtering Collaborative Filtering Matrix Factorization Text Analytics: Applications, Concepts, & Terminology Text EDA Vector Space Modeling: Set/Bag-of-Words TF-IDF Word Embedding Graphs Analytics: Major Classes of Graphs Network Descriptive, Connectivity and Centrality Analytics Large-Scale Network Structure Clustering and Communities Graph Representation Learning	Slides Slides Slides Slides Slides Slides
Data Visualization Visual Perception Gestalt Principles Context, Preattention Magnitude Estimation Visual Attributes and Mapping Evaluating Visualization	Slides
Information Retrieval (IR) Term-Document Incidence Matrix Inverted Index Ranked Retrieval	Slides
Linear Algebra Review Vector Operations Linear Functions Linear Transformation Change of Bases Eigen Value & Eigen Vectors Powers of Matrices Random Walk Markov Chain	Slides
Data Structures Review List, Stack, Queue, Set, Dictionary, Heaps	Slides
Web Search Web Searches Challenges Content Spamming Trustworthy Webpages and Node Centrality PageRank: Dangling Nodes, Spider Traps, Random Teleporting Link Spamming Personalized and Topic Sensitive Pagerank Hyperlink-Induced Topic Search	Slides
Spectral Clustering Limitations of Distance Based Clustering Graph Partition and Cuts Spectral Graph Theory (Un)Normalized Graph Laplacians Relation of Graph Laplacian and Partition Spectral Clustering into k-Clusters	Slides
Proximity Problems & Curse of Dimensionality Proximity Problems on High Dimensional Data Distance Matrix Computation k-Nearest Neighbor Problem Fixed-Radius Nearest Neighbors Approaches for kNN problem Curse of Dimensionality Processing and Storage Data Sparsity Issues for Nearest Neighbors Huge Search Space Diminishing volume of n-ball Nearest neighbor instability Distance Concentration Angle Concentration	Slides	Notes Notes
Data Preparation & Representation Learning Data Preparation Data Compression Low Distortion Embedding Dimensionality Reduction Multi-dimesnsinal Scaling Dimensionality Reduction Feature Selection and Extraction Johnson-Lindenstrauss Lemma	Slides	Notes
Locality Sensitive Hashing (LSH) LSH for k-NN and Near Duplicate Problems LSH for: Hamming Distance Jaccard Distance Cosine Distance Euclidean Distance Constructing New LSH Families Non-LSH-able Distance Measures Data Dependent LSH	Slides	Notes
Data Streams Data Stream: Model of Computation Synopsis Based Exact Stream Computation Sliding Window, Histogram and Wavelets Sketches: Count, Count-Min & AMS	Slides	Notes
Principal Component Analysis (PCA) Aims of PCA PCA vs. JL-Transform Variance-Covariance Matrix PCA Objective: Reconstruction Error and Projected Variance Linear Algebraic Formulation Eigen Decomposition of Covariance Matrix Power Iteration Method Eigenfaces Limitation of PCA	Slides	Notes
Matrix Factorization and SVD Rank Factorization of a Matrix Low Rank Approximation Singular Value Decomposition (SVD) SVD Applications: Recommendation System Latent Semantic Analysis Data Denoising PCA and SVD	Slides	Notes

Back to Top