Inventory forecasting tool for atrading company
A tool that forecasts stock requirements from sales history, so purchasing stops being guesswork.
- Client
- Meridian
- Type
- Platform
- Year
- 2024
- Delivered
- Platform build · Data analysis · Forecasting · Dashboard
- Type
- Forecasting tool
- Output
- Recommendation + range
- Review
- Forecast vs actual
- Data
- Consolidated history
01 /
The challenge
Sales history was scattered across systems and formats, and demand was seasonal — a simple average would overstock the quiet months and understock the peaks. The team also needed to trust the forecast before they would act on it.
02 /
What we did
- 01
Consolidated sales history from the existing systems into one clean dataset the forecast could actually use.
- 02
Built a forecasting model that accounts for seasonality and trend, so the recommendation reflects how demand actually behaves.
- 03
Designed the output as a purchasing recommendation with a confidence range, not a single number, so the team can apply judgement.
- 04
Added a dashboard comparing forecast against actuals, so the team can see where the model is right and where it needs adjustment.
03 /
Outcome
Purchasing decisions that start from data rather than memory. The team can see what is likely to be needed, when, and how confident the forecast is — and the comparison against actuals means the model improves as it is used.
Built with
- Python
- React
- PostgreSQL
- Pandas