Philip Goddard

13 April 2016

Coursera Machine Learning Specialisation: Part 1

I recently signed up to the machine learning specialisation offered by Washington University. As it is a six-course specialisation, I figured I would write a post at the start with my initial thoughts, to be concluded in a few months time once I have completed it!

24 Mar 2016

Hello world- the genetic way

I have been meaning to get some practice with object-oriented programming in a language a little more 'usual' than R. I have dabbled with C++ and Python in the past, and as I have recently started using Python for data analysis, I figured it was a good choice to do some experimenting with. I've recently been investigating optimisation methods for automatic feature selection, so to kill two birds with one stone, here is a simple genetic algorithm written in Python.

16 Mar 2015

Ensemble Models: Various Flavours of Stacking

After reading a couple of articles on the subject of stacking predictions from regression models to gain an improved prediction, I decided to investigate for myself.

1 March 2015

Coursera Machine Learning

After gaining a lot of practice with 'applied' predictive modelling, I decided it was time to focus on getting my hands dirty and start to prototype my own learning algorithms to better understand what I was doing. This course had been on my to-do list for a while, and is probably the best massive open online course I have taken to date.

26 Feb 2015

Coursera Managing Big Data with MySQL

This course is probably the best resource I have used to learn how to query SQL databases. Personally, I would much prefer to be tinkering with a learning algorithm than writing SQL queries; however it is a vital skill to have in your toolkit if you are interested in data analysis. The only way you can learn is with practice, and you will get a hell of a lot of that with this course!

19 Feb 2016

Integrating custom models into the caret framework

I recently wrote my own neural network classifer in R. Time to find out how to integrate it with the caret framework!

10 Feb 2015

Home-brewing nnePtR: a neural network classifier

I have really been enjoying Andrew Ng's machine learning course on Coursera. The course demonstrates how to prototype a simple neural network classifier in GNU Octave, so I decided to extend these ideas in an R package.

27 Jan 2015

Churning with Caret: Trees and Rules

Part three of predictive modeling on the churn data set. This time I will have a look at tree and rule based models, and see how they stack up against linear and non linear models!

24 Jan 2015

Visualising Classification Boundaries

A quick, informative way to visualise classification boundaries.

21 Jan 2015

Churning with Caret: Non-linear Models

Part two of predictive modeling on the churn data set with nonlinear models. I've already tackled linear models, and demonstrated that they leave a lot of room to be improved on. Let's see how non-linear models rise to the challenge!

<123 4 >