BITS WILP Data Science Knowledge Base: Semester 1 Resources

EMDS (known as MFML in the AIML program) is widely regarded as the most challenging course in Semester 1. It establishes the mathematical language of data science: Vector Spaces, Rank-Nullity theorem, Basis, Reduced Row Echelon Form (RREF), Eigenvalues & Eigenvectors, Cholesky and Singular Value Decomposition (SVD), Taylor approximations, Jacobians, Hessians, and Constrained Optimization (KKT conditions). This course cannot be crammed the night before the exam.

PFARD focuses on transforming raw, dirty enterprise data into curated, model-ready inputs across diverse modalities (audio, text, image, graph, spatial, and time-series). Key modules cover data wrangling (munging), contextual enrichment of metadata, multimodal feature extraction, knowledge graphs, pre-processing for deep learning, and responsible AI governance (fairness, explainability, security, and regulatory compliance).

Introduction to Data Science provides the analytical core of the program: Data and Data Models, Data Wrangling, Feature Selection, Supervised Classification & Prediction (Decision Trees, Naive Bayes, Logistic Regression), Association Rule Mining (Apriori, FP-Growth), Clustering (K-Means, DBSCAN, Hierarchical), Anomaly Detection, and Exploratory/Explanatory Data Analysis with visual storytelling and ethics in AI.

Modern Database Systems investigates distributed data architectures: Foundations of Structured Data Management (SQL & Relational Algebra), Distributed Storage (Sharding, Partitioning, Replication), CAP Theorem Trade-offs, ACID vs. BASE Consistency Models, NoSQL Stores (Key-Value, Document, Column-Family, Graph), Cloud Databases & NewSQL, Data Warehouses, Data Lakes, Lakehouses (Delta Lake, Iceberg), and Specialized Databases (Vector DBs for GenAI, Streaming, Spatial).