AI Blocks › Track your numbers
Quick time series forecaster
Give it a CSV of past values. Get a forecast with ranges and a test on held-back data, with no model training.
Quick time series forecaster is a free AI skill for the job "Track your numbers". Give the instructions to an AI assistant that supports the required files and tools, then review its output.
Free · No account · MIT licence · Made by K-Dense · Thinking: low
Made by K-Dense. See the original on GitHub; we copied it as it is, with its licence.
What it does
A forecast of your numbers with ranges around it, tested against data you held back, and no model training. You give it a CSV of past values and see how well it would have predicted what you already know. The skill cleans the file into a regular time grid, then runs Google's TimesFM, a pre-trained forecasting model, that predicts without fitting to your data. You get a middle estimate plus low and high bands. It scores the forecast on the held-back period so you can judge trust before relying on it. It is not a tool for cause and effect. It needs Python, PyTorch and a one-time model download.
You give it
- CSV of past values with dates
- How far ahead to forecast
- Python with PyTorch installed
- Extra driver columns (optional)
You get
- Forecast with low and high ranges
- Score on held-back data
- Cleaned time series
- Chart of past and predicted values
Use it when
- You have monthly sales and want the next six months.
- You need a forecast but have no time to build a model.
- You want to see how reliable a forecast is before using it.
- Your data has gaps or uneven dates that need tidying first.
How to use it
- Add it to your AI once: download it and add the folder to Claude, ChatGPT or Codex as a skill.
- Ask for the job in your own words. Your AI opens the skill when it fits.
- It works best when your AI can run code and download a model.