AI Specialization Program

• Soft AI Academy
AI Specialization Program
•
Technical Tracks
AI Engineering & Data
•
Week 3

WEEK 3 —

Machine Learning Foundations

Theme: Build & Evaluate Models Properly

Week 3 is about:

Understanding classical Machine Learning deeply

Before touching deep neural networks.

Week 3 —

6 Lesson Structure

Lesson 1:

Supervised vs Unsupervised Learning

Focus:

What is supervised learning?

What is unsupervised learning?

Labels vs no labels

Real-world use cases

Choosing the right approach

Outcome: Students understand ML problem types and task framing.

Lesson 2:

Regression Models

Focus:

Linear regression intuition

Cost functions

Gradient descent concept

Evaluation metrics (MAE, MSE, RMSE)

Practical implementation in Python

Outcome: Students can build and evaluate regression models.

Lesson 3:

Classification Models

Focus:

Logistic regression intuition

Decision trees basics

k-Nearest Neighbors intuition

Confusion matrix

Precision, Recall, F1-score

Outcome: Students can build and evaluate classification systems.

Lesson 4:

Model Evaluation & Overfitting

Focus:

Train vs validation performance

Overfitting vs underfitting

Bias–variance tradeoff revisited

Cross-validation

Model comparison strategies

Outcome: Students can diagnose model performance issues.

Lesson 5:

Feature Selection & Model Improvement

Focus:

Feature importance

Removing irrelevant features

Hyperparameter tuning concept

Regularization intuition

Improving generalization

Outcome: Students can improve weak models systematically.

Lesson 6:

ML Project Integration (Mini Project)

Focus:

End-to-end small ML project

Dataset → clean → train → evaluate

Proper experiment documentation

Basic reporting

Outcome: Students complete first structured ML project.

Week 3 Final Outcome

By the end of Week 3, students can:

✔ Clearly distinguish supervised vs unsupervised problems

✔ Build regression models from scratch

✔ Build classification models

✔ Evaluate models using proper metrics (MSE, Precision, Recall, F1, etc.)

✔ Diagnose overfitting and underfitting

✔ Apply cross-validation properly

✔ Improve models using feature selection & tuning

✔ Complete an end-to-end classical ML mini project.

Week 3 produces:

Competent Machine Learning practitioners.