Learning By Failure | Machine Learning

A repository of sample notebooks to learn concepts of Applied Machine Learning in a Project Oriented Approach.

The repository is a part of the Learning By Failure series, where we learn concepts by failing and understanding the reasons behind the failure.

We use a central theme project to first build a need to learn the concepts of Machine Learning and then apply them to solve the problem.

Please note: Stock Market is a complex domain and the project is only for educational purposes. The project is not intended to be used for real-world trading. The "Learning By Failure" concept is intended to target something so aspirational that we wish learn more to try and achieve our goals.

Below is the outline of the course:

• Module 1: Introduction
  • Introduction to the World of AI/ML
  • Machine Learning and Its Applications
  • Types of Machine Learning:
    • Matrix of problems: Supervised/Unsupervised, Regression/Classification
  • Introduction to Case Study/Project for Project Oriented Approach: Stock Market
• Module 2: Defining the Problem Statement
  • Understanding and Framing the Problem
  • Intelligence vs Inference: understanding what to expect
  • Setting up Expectations and Evaluation Metrics
  • Introduction to Jupyter Notebooks
  • Resource:
    • Downloading Data
    • Python Library pre-requisites
    • Notebook Repository
• Module 3: Knowing your data
  • Describing Data
  • Visualizing Data
  • Preprocessing Data
    • Scaling
    • Null Handling
    • Outliers
• Module 4: Supervised Learning
  • Defining Classification and Regression
  • Simple Regression Model
  • Simple Classification Model
  • Inference
  • Evaluation Metrics
  • Running Backtests
• Module 5: Unsupervised Learning
  • Understanding clustering
  • Simple clustering Model
  • Inference
  • Evaluation
  • Running Backtests
• Module 6: Deep Learning
  • Introduction: Neuron
  • Neural Networks
    • ANN
    • CNN
    • RNN
  • Training a Simple Model
  • Running Backtests
• Module 7: Reinforcement Learning
  • Introduction
  • Environments, Agents, Rewards: Stable Baselines, OpenGym
  • Defining Environment
  • Understanding and Designing Rewards
  • Training a Basic Model
  • Running Backtests