Machine-Learning-Lab

This repository contains a series of machine learning projects from the Programming Machine Learning lab course, which is part of the Data Analytics Master's program at Hildesheim Universität.

Content:

1. Introduction
   • Word counter program
   • Image blurring program
2. Linear Regression
   • Exploratory analysis of Rossman GmbH sales data
   • Linear Regression through normal equations
   • Multiple Linear Regression (MLR)
   • Multivariate Multiple Regression on Rossman GmbH sales data
3. Gradient Descent
   • Gradient descent on Rosenbrock function
   • Preprocessing of 3 real-world datasets: Airfare and demand, Wine Quality, Parkisons Dataset
   • Linear Regression with gradient descent
   • Step length control for gradient descent
     • Backtracking
     • Boldriver
     • Look-head optimizer
4. Logistic regression
   • Preprocessing of tic-tac-toe dataset
   • Logistic regression with gradient ascent
   • Logistic regression with Newton's method
5. Variable selection, regularization and hyperparameter tuning
   • Preprocessing of Bank marketing dataset
   • Logistic regression with mini-batch gradient ascent
   • Backward search for variable selection
   • Regularization for logistic regression
   • Hyperparameter tuning through grid search with k-fold cross-validation
   • Hyperparameter tuning through Hyperband
6. Polynomial regression
   • Preprocessing of Wine Quality dataset
   • Regularized linear regression
   • Hyperparameter tuning through grid search
   • Polynomial regression with Sckit learn
7. K Nearest Neighbors
   • Preprocessing and EDA of UCR Time Series datasets
   • Dataset Imputation with KNN
   • Classification with KNN
   • KKN acceleration techniques
     • Partial Distances
     • Locality Sensitive Hashing
8. Neural Networks
   • Optical Character Recognition on MNIST dataset
   • Hyperparameter tuning through random search
   • Car steering angle prediction via CNN
9. Decision Trees
   • Classification with decision tree on Iris dataset and Car Evaluation dataset
     • Misclassification Rate as quality-criterion
     • Information gain as quality-criterion
   • Gradient Boosted decision tree
10. Matrix factorization
    • Exploratory analysis of movielens 100k dataset
    • Matrix factorization for recommendation
11. Naïve Bayes and SVMs
    • Preprocessing of 20newsgroups dataset
    • Naive Bayes classifier
    • SVM classifier