Machine-Learning-Specialization-University of Washington

Programming Assignments for machine learning specialization courses from University of Washington through Coursera.

Techniques used: Python, pandas, numpy,scikit-learn, graphlab, R

In terms of the library and packages, I only used graphlab and SFrame for Machine Learning Foundations. For all the other courses (Regression, Classification and Clustering) I have used pandas for feature enginering and scikit-learn to build out modeling.

Specialization Courses:

• Machine Learning Foundations: A Case Study Approach

  Regression: Predicting House Prices (Leverage Zillow data to build linear regression model to predict house prices)

  Classification: Analyzing Sentiment (Build logistic classification model to analyze product sentiment)

  Clustering and Similarity: Retrieving Documents (conduct cluster analysis for document retreival, tf-dif)

  Recommending Products: Build Matrix Factorization Model and leverage Jaccard Similarity to Recommend Songs

• Machine Learning: Regression

  Project Overview: How to predict a house's price? How to evaluate model? How to prevent model from overfitting?

  Simple Linear Regression: Implementing closed-form solution for simple linear regression

  Multiple Linear Regression: Exploring multiple regression models for house prediction; Implementing gradient descent for multiple regression

  Assessing Performance

  Ridge regression

  Lasso regression

  Kernal regression

• Machine Learning: Classification

  Project 1 Overview: Build classification modeling to predict if an Amazon review is positive.

  Project 2 Overview: Is this loan safe or risky?

  In these assignments, I have built logistic regression modeling and decision tree modeling to predict if a loan is risky or safe and test classification errors for different models by both using scikit-learn and implementing the (greedy ascent, greedy descrsion tree and etc.) algorithm from sracth.

  Linear Classifiers & Logistic Regression

  Learning Classifiers; Overfitting & Regularization in Logistic Regression

  Decision Trees

  Precision-Recall

  Stochastic Gradient Ascent

  SVM http://www.svm-tutorial.com/2014/11/svm-understanding-math-part-2/

• Machine Learning: Clustering & Retrievel

  Nearest Neighbor Search

  Clustering with K-Means

  Mixture Models (Implementing Expectation Maximization Algorithm for Gaussian mixtures; Clustering text data with Gaussian mixtures)

  Mixed Membership Modeling via Latent Dirichlet Allocation

Others:

computational cost (comlexity) http://stackoverflow.com/questions/2307283/what-does-olog-n-mean-exactly

bitwiseoperators ( 0 1 ) https://wiki.python.org/moin/BitwiseOperators

additional blog that helps understand LDA http://confusedlanguagetech.blogspot.com/