Machine Learning

Introduction

These is a culminations of projects I worked on to complete the machine learning course by Andrew Ng on Coursera.

Projects

Linear Regression

Created a linear regression algorithm to predict profits for a food truck. The data that was given was profits and population of cities where the food truck was located. We will use this knowledge to help us decide where to open a new food truck location.

This is the data we have to sort through to find a correlation to solve the problem

The line drawn through the data then allows us the user to determine where to next open a food truck.

Logistic Regression

Implemented regularized logistic regression to predict whether microchips from a fabrication plant passes quality assurance (QA). During QA, each microchip goes through various tests to ensure it is functioning correctly. Have the test results for some microchips on two different tests. From these two tests, determine whether the microchips should be accepted or rejected. To help you make the decision, you have a dataset of test results on past microchips.

This the data of past microchips that passed or failed failed the test that we have to sort through.

After running through logistic regression we are able to create this boundary curve that is a good predictor of deciding if a microchip will either pass or fail the tests.

Multi-class Classification

Programmed a multi-class classification system to read hand written numbers and determine with high accuracy the number to help in determining address for the USPS.

The data set of handwritten numbers from which we will test the accuracy of our algorithm.

The resulting accuracy of the algorithm.

Neural Networks

Instead of using Linear Regression to help determine the numbers I created a Neural Network to help improve the effectiveness of the algorithm.

The data set of handwritten numbers from which we will test the accuracy of our algorithm.

This is how the neural network is setup with one input node, one hidden node, and one output node.

This is how the neural network determine to see what the number is after running / testing the algorithm.

The resulting accuracy of the algorithm.

Regularized Linear Regression

Implemented regularized linear regression to predict the amount of water flowing out of a dam using the change of water level in a reservoir.

The data of water levels that we have to fit a polynomial line through.

The line of best fit through the data we are given to help us predict other cases.

Support Vector Machines

Utilized a support vector machines (SVMs) to build a spam classifier. This is the type of algorithm that is used in services like gmail, outlook, or etc. These emails are then put in your spam inbox so you don’t have them clutter your inbox.

These are the data points that the algorithm looks for in the data to see if it is a spam email or not.

The output of a test case email to see if it actually marks it as spam.

Recommender Systems

Created an algorithm similar to recommendation systems used in Netflix or YouTube to help show you videos that might interest you.

The output of the recommender system.

Certification

A picture of the certificate I achieved after completing the course on Coursera.