Applied Data Science - Live

Class 1: Getting Started with Python

• Install Python; set up Jupyter, Colab, and Kaggle

• Learn basic syntax: Variables, data types, loops, conditionals, and error handling

• Introduction to GitHub and version control

Class 2: Python Data Structures

• Work with lists, tuples, dictionaries, and sets

• Write functions (including lambda expressions) and perform file I/O (CSV, text)

• Practice exercises on manipulating lists and dictionaries

Class 1: NumPy & Pandas

• Create and manipulate NumPy arrays: Slicing, vectorization, and broadcasting

• Use Pandas for merging, cleaning, handling missing data, and performing descriptive statistics

• Reference: The NumPy Array: A Structure for Efficient Numerical Computation (van der Walt et al., 2011)

Class 2: Data Plotting & Simple Transformations

• Plot data using Matplotlib and Seaborn (line, bar, scatter, and histogram plots)

• Apply basic feature transformations: Scaling (Standard/MinMax) and encoding (one-hot, label)

• Project: DataViz Explorer - Clean and visualize a dataset

Class 1: Feature Engineering

• Understand why transforming raw data is important

• Apply key techniques: Log transform, binning, and polynomial features (with simple math)

• Encode categorical data using one-hot, label, and target encoding

• Project: Feature Mastery - Implement and compare feature transformation techniques

Class 2: Building an ML Pipeline

• Overview of supervised vs. unsupervised learning

• Learn the steps in the ML workflow: Preprocessing, training, validation, and testing

• Understand data splitting methods and the rationale behind cross-validation

• Performance Metrics:

  • Classification: Accuracy, Precision, Recall, F1 Score, ROC AUC, Confusion Matrix

  • Regression: MSE, RMSE, MAE, R Square, adjusted R Square

• Project: ML Basics Pipeline - Build a simple pipeline on a toy dataset and evaluate its performance

Class 1: Linear Regression

• Derive the least squares solution and MSE cost function

• Explain gradient descent: Derivatives, update rules, and learning rate

• Project: Linear Predictor - Code linear regression from scratch and compare results with scikit-learn; evaluate using MSE, RMSE, and R Square

• Reference: Learning Representations by Back-Propagating Errors (Rumelhart et al., 1986)

Class 2: Logistic Regression

• Understand the sigmoid function and binary cross-entropy loss (with mathematical explanation)

• Project: Binary Classifier - Implement logistic regression (from scratch and via scikit-learn); evaluate using confusion matrix, accuracy, precision, and recall