An overview of how SHAP feature contributions are calculated — Suppose you (Player 1) and a friend (Player 2) enter a Kaggle contest. You end up winning the first prize of $10,000. Now, you want to split the money fairly. Your friend suggests that you just split it equally. However, your hyperparameter tuning skills are superior. You believe you deserve…

The intuition, maths and code (R and Python) behind partial dependence plots and individual conditional expectation plots — Both PDPs and ICE plots can help us understand how our models make predictions. Using PDPs we can visualise the relationship between model features and the target variable. They can tell us if a relationship is linear, non-linear or if there is no relationship. Similarly, ICE Plots can be used…

Doing an exploratory fairness analysis and measuring fairness using equal opportunity, equalized odds and disparate impact — It is no longer enough to build models that make accurate predictions. We also need to make sure that those predictions are fair. Doing so will reduce the harm of biased predictions. As a result, you will go a long way in building trust in your AI systems. To correct…

My process, tools and tips for writing data science articles — I find the blank page of every new article to be a bit daunting. There is so much work to be done. You need to do research, draft, edit and add graphics. In some cases, you will need to learn something completely new. …

An introduction to IML — the field aimed at making machine learning models understandable to humans — Should we always trust a model that performs well? A model could reject your application for a mortgage or diagnose you with cancer. These decisions have consequences. Serious consequences. Even if they are correct, we would expect an explanation. A human could give one. A human would be able to…

The differences in culture, workload, mentorship, learning rate and access to data and tools — Walking into the office looked more like a day at the beach. Shorts and flip-flops. Was this startup culture or just Cape Town? From the CEO to the new hire (me), the entire 8-person team worked from the same room. During my time there, I briefly consulted for one of…

Using a heatmap to visualise a confusion matrix, time-series movements, temperature changes, correlation matrix and SHAP interaction values — Heatmaps can bring your data to life. Versatile and eye-catching. There are many situations where they can highlight important relationships in your data. Specifically, we will discuss how you can use them to visualise: A confusion matrix for model accuracy

Why my first job in data science was not what I expected — New shirt, new shoes. I was ready for my first job in one of Ireland’s biggest banks. I was excited. Looking back, I had good reason to be. I was able to work on impactful projects and I learned an immense amount. In fact, the biggest lesson was: Data science…

Avoid post-processing the SHAP values of categorical features — Combining CatBoost and SHAP can provide powerful insight. Especially, when you are working with categorical features. The way CatBoost handles these features makes it easier to understand your model with SHAP. With other modelling packages, we need to first transform categorical features using one-hot encodings. The problem is that each…