This repo contains my submissions for the coursera course on Machine learning categorized by the week.

This course covers topics like:

• Linear regression to build a prediction model with a training set.
• The concept of a cost function.
• Logistic regression to build a prediction model for clasiffication problems. -- Using sigmoid functions to restrict the output range to 0 - 1 for classification problems.
• The concept of underfitting, overfitting, regularization to avoid overfitting.
• Using neural networks to improve the performance of building the prediction model when higher order polynomials are in play.
• Definition of sensitivity and its importance in finding the optimal theta values.
• Using backward progaration to find the gradients. Using gradient checking to verify your implementation of backward propagation.
• Using gradient descent or more optimal algorithms such fminuc to find the optimal theta.
• Using a cross-validation and test set to pick the right model for machine learning.
• Large margin classifiers, Support vector machines and Kernels.
• Definition and importance of Precision, Recall and F1 score.
• Anomaly detection using mean, variance and co-variance.
• Building a recommendation engine using collaborative filtering.

• Ubuntu 14 or later.
• sudo apt-get install octave
• sudo apt-get install git
• git clone https://github.com/pkrishn6/MachineLearning.git
• cd into any of the weekly submissions and open ex.pdf to understand the goals of the assignment.
• cd into the weekly submissions from octave cli and run ex to see the code in action.

• https://www.coursera.org/learn/machine-learning/home/welcome
• http://numericinsight.com/uploads/A_Gentle_Introduction_to_Backpropagation.pdf
• http://stackoverflow.com/questions/2480650/role-of-bias-in-neural-networks
• http://outlace.com/Computational-Graph/