Context parroting: A simple but tough-to-beat baseline for foundation models in scientific machine learning https://arxiv.org/abs/2505.11349

Context parroting relies on short stretches of time-series data (or context). As it moves through the time series, it scans for similar patterns or motifs that appeared earlier in the sequence, and uses those patterns to predict what might come

https://openreview.net/forum?id=EUAXc9Hlvm

https://www.santafe.edu/news-center/news/a-simple-baseline-for-ai-forecasting-in-machine-learning

#machineLearning #forecasting #timeseries #forecasting #ML

Anomaly detection isn't just for metrics.

#VictoriaMetrics Anomaly Detection supports additional input data sources through #VictoriaLogs reader, allowing you to monitor #log-derived and traces-derived metrics for anomalies.
This expands the versatility, enabling it to handle a wider range of data sources beyond #timeseries metrics from VictoriaMetrics or #Prometheus, including VictoriaLogs and #VictoriaTraces.

🕐 2026-04-04 12:04 UTC

📰 パラメータ4個で710M超えのFoundation Modelに勝った時系列予測手法FLAIRの全貌 (👍 35)

🇬🇧 FLAIR beats Amazon's 710M-parameter Chronos model with just 4 parameters. No GPU needed, pure numpy/scipy. Simple yet powerful time series forecast...
🇰🇷 FLAIR가 단 4개 파라미터로 Amazon의 7억1천만 파라미터 Chronos 모델을 능가. GPU 불필요, numpy/scipy만으로 구현된 시계열 예측 기법.

🔗 https://zenn.dev/t_honda/articles/flair-time-series-forecasting

#TimeSeries #MachineLearning #Zenn

🚀 New #rstats 📦: ecoXCorr now available on CRAN!

https://cran.r-project.org/package=ecoXCorr

It provides a simple workflow to explore lagged associations between environmental #timeseries and eco / epidemio responses:
➡️ flexible lag intervals
➡️ GLMM via glmmTMB
➡️ plot cross-correlation maps

see REDME: https://github.com/Nmoiroux/ecoXCorr

🌐🧪🌍

#ecology

Working with time-series data at scale? “How Prometheus Keeps Its TSDB Sane” breaks down how Prometheus keeps its own storage manageable and safe.

Read More: https://zalt.me/blog/2026/04/prometheus-tsdb-sanity

#Prometheus #TSDB #timeseries #observability