ml-zoomcamp

This repository contains exericises and notes I took taking the ML Zoomcamp provided by https://github.com/alexeygrigorev/mlbookcamp-code

Week 1

• Basics of Python: Pandas and Numpy

Week 2

Linear Regression: Car Price Prediction

• Data Preparation
• Exploritoy Data Analysis
• Setup the Validation Framework
• Linear Regression
• Linear Regression Vector Form
• Traiining a Linear Regression Model
• Car Price Baseline Model
• RMSE
• Validating the Model
• Simple Feature Engineering
• Categorical Variables
• Regularization
• Tuning the Model
• Using the Model

Week 3

Logistic Regression for Classification: Churn Prediction

• Read Data and initial Data Preparation
• Setting up the Validation Framework
• EDA
• Feature Importance: Churn Rate and Risk Ratio
• Feature Importance: Mutual Information
• Feature Importance: Correlation
• One-Hot Encoding
• Logistic Regression
• Training Logistic Regression with Scikit-Learn
• Model Interpretation
• Using the Model
• Summary
• Explore more

Week 4

Evaluation Metrics for Classification

• Read Data and Data Preparation
• Model Setup
• Make Predictions
• Accuracy: Evaluate a Dummy Model
• Confusion Table
• Precission and Recall
• ROC Curves
• ROC AUC
• Cross Validation
• Summary
• Explore more

Week 5

Deploying Machine Learning Models

• Read and prepare the Data
• Train the Model
• Save and Load the Model
• Web Services: Introduction to flask
• Serving the churn model with flask
• Dependency and Environment Management: Pipenv
• Environment Management: Docker
• Deploying in the Cloud
• Summary
• Explore more

Week 6

Decision Trees and Ensemble Learning: Credit Risk Scoring

• Data Cleaning and Preprocessing
• Decision Trees: sklearn
• Decision Trees: explained
• Decision Trees: Parameter Tuning
• Ensembles and Random Forests
• Gradient Boosting and XGBoost
• XGBoost: Paramter Tuning
• Selecting the final Model
• Summary
• Explore more

Week 7

Midterm Project

Week 8

Deep Learning: Fashion Classification

• Introduction to Tensorflow and Keras
• Pretrained CNN
• CNNs
• Transfer Learning
• Hyperparamter Tuning: Adjusting the Learning Rate
• Checkpointing
• Add more Layers
• Regularization and Dropout
• Data Augmentation
• Train a larger Model
• Using the Model
• Summary

Week 9

Serverless Deep Learning

• Overview
• AWS Lambda
• Tensorflow Lite: Convert Tensorflow Model to Tensorflow Lite
• Prepare the lambda Code
• Creating a lambda Function
• API Gateway: Exposing the Lambda Function

Week 10

Tendorflow Serving

• Overview
• Tensorflow Serving: Convert model to saved_model format
• Create a Preprocessing Service
• Run everything locally with Docker-Compose
• Introduction to Kubernetes
• Deploy a simple service to Kubernetes
• Deploy the Tensorflow Model to Kubernetes
• Deploy to EKS
• Explore more