Skip to content

Machine Learning exercises for my subject of Machine Learning at University of Granada (UGR).

License

Notifications You must be signed in to change notification settings

BraulioV/Machine-Learning

Repository files navigation

Machine Learning

Machine Learning exercises for my subject of Machine Learning (Aprendizaje Automático) at University of Granada (UGR) of 2015/2016.

This repository contents Machine Learning exercises divided by sections, corresponding to the exercises of the subject. Each folder contents the R code and a R-markdown that you can use on RStudio and see the results of the exercises, edit the content, etc. Also, a PDF is included (in Spanish) with the results and conclusions of the differents exercises.


1. Perceptron and linear regression

Here, I implemented a linear perceptron, pocket perceptron and linear regression, and they are tested with random generated data to see how perceptrons and regressions works.

2. Linear methods, overfitting and regularization

Here, I implemented differents linear methods:

  • Gradient descent, also known as steepest descent.
  • Coordinate descent.
  • Newton method.
  • Logistic regression.

Also, there are implementations to measure the overfitting of the model, apply regularization to a model, weight decay, etc and model selection according to their perfomance.

3. Generalized linear models, SVM, KNN and Trees

In this section, I followed the book An Introduction to Statistical Learning with Applications in R (ISLR), and do different exercises with generalized linear models, Support Vector Machine, KNN, Boosting, Bagging, Random Forest and Trees... and different dataset.


Some data visualization are made with ggplot2, so, in order to execute the code, this package is required.