Let’s demand real comparisons and rigorous benchmarks.

Otherwise, we’re just building castles in the air.

#TimeSeries #Transformers #NBEATS #MLResearch #DataScience #IntegrityMatters

🔹 **N-HiTS**, a variant of the very successful forecasting model (NBEATS), enables hierarchical interpolation and multi-rate sampling, solving long-horizon forecasting challenges.

Gracias Spotify por incluir temarracos como este en mi descubrimiento semanal
open.spotify.com/track/1v5gM4...

Zayed University researchers used data from 4 countries and found that adding Google and Apple mobility data helped an N-BEATS deep learning model forecast new COVID-19 cases more accurately than LSTM Markov.

journals.plos.org/plosone/arti...

NBEATS* enhances the existing NBEATS on energy demand forecasts by having all blocks forecast/backcast standardized values (shape is important!) and custom loss functions

💸  the second benchmark that new model “won” against is a weak model DeepAR relying on normality assumptions and that has been shown to be outperformed by ARIMA and other classical models.

💸 Authors also excluded high performing non transformer architectures including NBeats and NHITs

#TimeSeries #Forecasting #MachineLearning #Transformers #LLMs #DataScience #NBeats #AI

Machine Learning-Based Characterization of Solar p-Mode Frequency Shifts during Solar Cycle 25. Rekha Jain (1) et. al. https://arxiv.org/abs/2604.20802