Do you really know how this hyperparameter works? — One of the most creative and mysterious aspects of the modelling work is the hyperparameter setting. It’s also where many people follow a random guide without really trying to understand the details. If you happen to be one of these people, let’s change that! In this post, we are going…

How to help your RStudio work with interactive apps — Shiny is a quite well-known and widely used R package that lets you develop interactive web apps in R, without the need for HTML, CSS, or JavaScript coding. I recently discovered for myself that Shiny has a much lesser-known feature, the so-called Shiny gadgets. …

A step-by-step recalculation in LightGBM and R, from data to predictions, using the banknotes dataset — A few weeks ago, I did a deep dive into how gradient boosted models work. …

Step-by-step recalculation of the hidden model results, using R and LightGBM — Gradient boosted models are often labelled as black box examples of machine learning algorithms. A black box model is defined on Wikipedia as a “system which can be viewed in terms of its inputs and outputs […] without any knowledge of its internal workings”. Is this actually an accurate depiction…

Three ways to move from GLM/GAM to a GBM model in insurance. — A common task during claim frequency modelling in an insurance setting is handling the different levels of exposure. Exposure is a broad term, Insurancepedia defines it as “susceptibility to losses or risks”, but in our example, it will be analogous with time. We expect higher claim frequency if an entity…

A case study in R about how being pedantic with your random seeds can lead to unwanted results. — Manually setting the seeds of a (pseudo) random generator is a double-edged sword: it does mean that running the same script will always yield the same results, thus making the experiment reproducible; but it might also take away more than what you intended from the randomness, thus making the experiment…

Except for this one of course. You can trust this one! Or can you? — I still remember the euphoria I felt when I first dived headfirst into the world of Data Science. You mean to tell me that all this information is just out there, available 24/7, completely free? From tutorials to code bits and complete packages, written by like-minded individuals, covering every possible…

With Examples in Python — I’m a big advocate of using simple models, whenever the problem allows them. It’s all fun and games with the fancy neural networks, but if the good old linear regression gets the job done, why not use it? However, linear regression being a seemingly simple concept, there are two issues…

The method I use to optimise the regularising parameters. A guide in Python for scikit-learn. —