Automated periodicity detection in time series

We recently presented a scientific paper during ECML PKDD Advanced Analytics and Learning on Temporal Data workshop part of the ECML-PKDD 2019 conference at Wurzburg, Germany.

24/09/2019

R&D

Tous les articles

Sommaire

Abstract

"A fully automated periodicity detection in time series" by Tom Puech, Matthieu Boussard, Anthony D’Amato, and Gaëtan Millerand.

To generate good models, you need the good features. For instance, knowing precisely the various periodicities of your temporal data is crucial. Does the time of day has a recurrent impact ? Does it exhibits another, fancier, periodicity ? This is precisely what this paper solves, by providing a new algorithm to automatically extract those features from your time series.

Download the file to read the full paper.

Une plateforme compatible avec tout l’écosystème

aws
Azure
Google Cloud
OVH Cloud
scikit-lean
PyTorch
Tensor Flow
XGBoost
jupyter
PC
Python
R
Rust
mongo DB