TimeGPT: The First Foundation Model for Time Series
Welcome to the 1st edition!
Welcome to the 1st edition of the AI Horizon Forecast newsletter!
We're here to break down milestone papers, extract valuable insights, and provide practical programming tutorials. Our focus: time series and the latest AI research.
Remarkably, August 2023 marked a critical milestone for the time-series research field with 2 major breakthroughs:
TimeGPT: Nixtla unveiled TimeGPT, the first foundation AI model for time-series forecasting.
AutoGluon–TimeSeries: Amazon published a study on AutoGluon–TimeSeries, an open-source library that generates accurate point & quantile forecasts with just 3 lines of Python code.
Let’s dive in!
Horizon AI Forecast is a reader-supported newsletter, featuring occasional bonus posts for supporters. Your support through a paid subscription would be greatly appreciated.
TimeGPT: The First Foundation Model for Time Series
In the first cohort of the “Modern Forecasting in Practice” course by top forecasting experts Tim Januschowski and Jan Gasthaus, the Nixtla team also participated.
During one of the sessions, as the conversation turned to the future of research in time series and modern large-scale architectures, Nixtla implied that public datasets were not large enough to test the potential of those larger architectures.
Their observation was spot-on.
That’s why Nixtla curated the largest collection of public time series data in history – over 100 billion rows of financial, weather, energy, and web data.