Back to Blog Späť na blog

Forecasting with QuantixAI

Forecasting with QuantixAI

QuantixAI turns raw time-series into clear forecasts you can explain and trust. In this walkthrough, you’ll create a baseline forecast, add what-if drivers, and learn how to read the outputs.

What you’ll get

  • Baseline forecasts using robust statistical + ML models
  • Scenario testing with exogenous features (e.g., marketing spend, holidays)
  • Optional hierarchical reconciliation so rollups match the parts

Step 1 — Connect and explore

Upload your time series (CSV/XLSX) or connect to your source. Use the visualization panel to confirm trend, seasonality, and outliers. If needed, apply transformations or filters.

Step 2 — Build a baseline forecast

  1. Open Forecast → select your series (or a group).
  2. Choose horizon (e.g., 12 weeks / 6 months).
  3. Let QuantixAI auto-select models (statistical + ML) or pick your own.
  4. Generate forecast to view the mean and the confidence interval.

Step 3 — Add what-if drivers

Include external features like campaigns, pricing, weather, or holidays. Adjust their future values to run scenarios (“+20% marketing spend in Q4”). QuantixAI recomputes the forecast with those assumptions.

Step 4 — (Optional) Hierarchical reconciliation

When you forecast across levels (SKU → Category → Region → Total), enable reconciliation so the sums are consistent at every level.

Interpreting results

  • Point forecast: your best estimate for each future period.
  • Intervals: expected range—wider bands = more uncertainty.
  • Feature impact (when using ML): how much each driver contributes to the forecast.

Tips

  • Start with a simple baseline, then layer drivers.
  • Compare scenarios side-by-side and export results for stakeholders.
  • Re-train periodically as data or drivers change.

Ready to try it on your data? Start your 14-day free trial—no credit card required.