Skip to content

Blog

ARMA is obsolete - but whats next?

It's a classic, around since the 70s, used it myself back in uni, but data science needs have clearly evolved.

With TSFresh, Blue Yonder's engineering team set the sails for brute-force feature engineering, combining a large set of 100+ predefined feature templates with traditional feature selection methods. This approach performs well, but their implementations limitations: computational complexity on real-world data and limited support for multivariate time series.

image

LLMs for relational data?

It's a classic, around since the 70s, used it myself back in uni, but data science needs have clearly evolved.

With TSFresh, Blue Yonder's engineering team set the sails for brute-force feature engineering, combining a large set of 100+ predefined feature templates with traditional feature selection methods. This approach performs well, but their implementations limitations: computational complexity on real-world data and limited support for multivariate time series.

image

Boosting Graph Neural Networks with getML: Automated Feature Engineering for Superior Model Performance

Graph Neural Networks (GNNs) excel at modeling complex data but do require feature engineering. This article shows how integrating getML's FastProp automates this process, boosting accuracy to 92.5% on the CORA datasetā€”surpassing the 90.16% benchmark. This approach simplifies implementation, reduces manual effort, and ensures consistent performance across a wide range of neural architectures, making it a valuable tool for data scientists for building consistent and high-performing models with minimal effort.