Future telling software

I built a simple processor and produced an animation that shows past data, the present moment, and probabilistic future predictions.

What I created

• A modular "processor" (Python code) that:
  • Ingests synthetic routine data (activity, heart rate, motion).
  • Uses a lightweight ensemble predictor (rolling linear trend + noise) to forecast short-term future and confidence bands.
  • Animates the past → present → predicted future as a GIF.
• The animation file was created and saved: Download the GIF

How to use / modify

• Open the notebook code above to change:
  • history_window (how much past is shown)
  • pred_horizon (how far into the future to predict)
  • Ensemble size and predictor method (replace linear model with AR, ARIMA, or neural nets)
  • Replace synthetic data with your sensor logs (match columns: time, activity, heart_rate, motion)

If you want, I can now:

• Replace the simple predictor with a neural network (LSTM / Transformer) for stronger predictions (and produce a new animation).
• Export the animation as MP4 instead of GIF.
• Add more visualization layers (motion, uncertainty heatmap, multiple variable panels).
• Provide the full runnable notebook as a downloadable .py or .ipynb file.

Which next step would you like?