Traditional forecasting treats over- and under-predictions equally, but in real-world scenarios like financial markets, the costs can differ significantly. Quantile forecasting addresses this by estimating confidence intervals, offering a nuanced view of uncertainty. In my latest post, I explore how this method works and cover its advantages.
MLflow is widely recognized as a powerful tool for tracking machine learning (ML) experiments, enabling data scientists and ML experts to systematically log metrics, parameters, and models. However, its capabilities extend beyond these traditional use cases — MLflow can also be used to track datasets.
I was offered a package of training programs covering several Microsoft Azure certificates. In a few posts, I’m going to explain each of these Azure certificates from the perspective of a data science and ML expert. The first post covers the Microsoft Azure Fundamentals certificate (AZ-900). I had access to official training by Microsoft and the learning material on learn.microsoft.com. Since the official learning material published on on learn.microsoft.com is the source for the exam, I will refer to the topics covered there.
I was asked to explore a car insurance premium dataset and predict a “fair” premium for each policyholder. The task included exploratory data analysis, feature engineering, model selection and comparison, and suggestions for model improvement. It was an interesting and insightful task, as our expectations sometimes fail and we can’t tackle the problem in a routine and straightforward way.
BentoML is an open-source platform for packaging ML models. It effectively simplifies the productionization of ML models by offering easy-to-use built-in features.