Recent blogposts
Don’t get fooled by integrated processes
In time-series analysis, you will frequently hear about the idea of integrated processes (e.g. I(1) or seasonal integration). The typical approach is differencing of your time-series in order to de-integrate it and make it stationary.
Multivariate, probabilistic time-series forecasting with LSTM and Gaussian Copula
You want to forecast a multivariate time-series using an autoregressive LSTM model. Since you need forecast intervals, you want to build a probabilistic model. The output of your model are the mean and standard deviation
Let’s make GARCH more flexible with Normalizing Flows
For financial time-series data, GARCH (Generalized AutoRegressive Conditional Heteroscedasticity) models play an important role. While forecasting mean returns is usually futile, stock volatility appears to be predictable, at least to some extent. However, standard GARCH
ARMA forecasting for non-Gaussian time-series data using Copulas
ARMA (AutoRegressive – Moving Average) models are arguably the most popular approach to time-series forecasting. Unfortunately, plain ARMA is made for Gaussian distributed data only. On the one hand, you can often still use ARMA
Why SMOTE is not necessarily the answer to your imbalanced dataset
When the topic of imbalanced classification comes up, many people praise SMOTE as the go-to approach. And indeed, popular algorithms have survived some evolutionary selection process by the community. However, we should still not just
If Your Models Are Underperforming, Build Better Datasets
Working with data can be hard. You might spend hours on your model or analysis without getting any reasonable results. At that point, it might be tempting to blame your performance issues on the wrong