# The Neuron

A neuron is a single node within a neural network.  By analogy with neurons within the brain we can think of a neuron “firing” in response to an input trigger, and we can think of machine learning as the process of training the neuron to recognise that input trigger.

(more…)

# Prepping Data

Most of the work associated with building a predictive model is associated with either performance tuning or data prepping.

I’m almost half way through prepping some data.  It’s not necessary to script this but a script allows me to adjust the data preparation in the future and more importantly to  document the sequence of steps that I have taken.

(more…)

# A Trivial Neural Network

I have been investigating the use of logistic regression to model image pixel data.  Now I want to take a look at the use of neural networks.  In this post I am going to build the simplest possible neural network and compare it against a simple logistic regression.

(more…)

# Logistic Regression pt. 2

In my last post I built a regression model with a single predictor variable.  That variable represented the value of a single pixel from a 28×28 image of a hand written digit.  In this post I will look at some model variations based on using a larger number of input variables.

(more…)

# Logistic Regression pt.1

In a recent post I created a table that contained two classes of data: images that represent either the handwritten digit ‘5’ or the digit ‘6’.  In this post I’ll model the data using logistic regression.  I will also take the opportunity to look at the role of training and test datasets, and to highlight the distinction between testing and validation.

(more…)

# Fives and Sixes

In my last post I was able to successfully re-orient a set of pixel data to reconstruct images of handwritten digits.  SInce version 12 of JMP we have been able to create expression table columns that can contain images.  That’s a logical location to store my newly revealed images:

(more…)

# Performance Profiling

In my last post I illustrated the performance boost generated by using matrix operations to conduct least squares regression calculations.  Matrices by their nature require numerical data.  So what about handling a categorical predictor variable?  To do this it’s necessary to create dummy variables – separate variables for each unique level of the predictor variable.

(more…)

# Regression in Matrix Form

I’m working on some predictive modelling projects and I need to iteratively compute R2 statistics over 100’s of variables.  Each time I do the calculations I need to go and have an extended coffee break – and I’m starting to buzz with too much caffeine so I thought I would look to see whether I could make my code more efficient!

(more…)