| Data Science Foundations |
Generalization, overfitting, validation strategy, metrics, hyperparameters, and a practical modelling workflow |
Open |
| Data Preprocessing |
Scaling, encoding, pipelines and information leakage, feature selection and engineering, missing values, imbalanced data |
Open |
| Fundamentals of Deep Learning |
The neuron, layers, training by backpropagation, optimizers, and the main network architectures |
Open |
| Practical Deep Learning |
PyTorch implementation, data handling and augmentation, transfer learning, dropout, batch norm, reproducibility |
Open |
| Lab 1: Binary Classification |
Predict campus placement: logistic regression from scratch, then kNN, SVM, trees, random forest, XGBoost and CatBoost |
Notebook Colab |
| Lab 2: Multi-Class Classification |
Classify air quality as Good, Moderate, Poor or Hazardous, from EDA and preprocessing to model comparison |
Notebook Colab |
| Lab 3: EDA and Data Preprocessing |
Predict used car prices: explore, clean and prepare a real-world tabular dataset |
Notebook Colab |
| Lab 4: PyTorch Basics |
Tensors, shapes and operations: the building blocks every later lab uses |
Notebook Colab |
| Lab 5: Neural Network in NumPy |
Build and train a two-layer network on MNIST from scratch, implementing every step yourself |
Notebook Colab |
| Lab 6: Neural Network in PyTorch |
The same MNIST classifier in PyTorch, with softmax and batch gradient descent |
Notebook Colab |
| Optional: Classification (Advanced) |
A second full classification workflow on weather-type data, from EDA questions to boosted trees |
Notebook Colab |
| Optional: R Squared |
R² and adjusted R² on a simple linear regression |
Notebook Colab |
| Optional: Images with PyTorch |
A complete MNIST pipeline: data preparation, model class, training and validation loops |
Notebook Colab |
| Optional: Tabular Data with PyTorch |
A neural network for weather-type classification on tabular data |
Notebook Colab |
