Scott C. Lowe

My doctorate studies are finally completed! The work was done, the thesis submitted, viva voce (oral examination) passed, and minor corrections completed, and today is the day of my graduation at Edinburgh. My PhD thesis is titled “Decoding information from neural populations in the visual cortex”. If you are interested... [Read More]

In this post, I’ll describe a technique for doing stratified partitions of datasets when your goal is regression instead of classification. Setting the scene Let’s say you’re training a model on a dataset and you need to split it into train and test partitions. You’re doing this so your... [Read More]

In this post, I’ll discuss feature scaling and an idea I had about rescaling features so they are transformed into Gaussian distributions in additon to standardisation. I’ll be doing the Gaussianisation using the Box-Cox transformation. But first, some background. The need to normalise features Normalising continuous scaled... [Read More]

In the past, whenever I’m working with a dataset which contains ordinal features, I’ve never known what to do with them. But now, I think I’ve come up with a good solution, which is the rank-hot encoder. Background Continuous data If you’re working with continuous valued fields with... [Read More]

Quite a while ago now, I was in the situation where I was moving from one MATLAB installation to another, and I wanted to transfer my custom-made GUI colour scheme settings from the old installation to the new one. Since I wanted a dark-background theme, I had long-ago customised the... [Read More]