The situation
The manufacturer needed accurate forecasts of sales volumes to adapt production capacity to market shifts.
Forecasting was built mainly in Excel, with indicators chosen by correlation, many of which had no causal effect on orders.
The challenge
- Minimise the forecast errorIrrelevant indicators gave poor accuracy.
- Find the indicators that matterIdentify the leading indicators that actually drive orders.
- Detect shifts and policiesSee market trend shifts early, and evaluate new policies and their impact on the aviation market.
What Indicio did
- Used a Lasso model to find the group of leading indicators with the highest impact on forecasted orders, and how their ranking changes over time.
- Included known future events, such as the Paris Air Show, and upcoming regulation similar to earlier policies, in the models.
- Weighted a large number of econometric models into one forecast.
The result
- 40%
- better market forecast accuracy
- 1–2 months
- earlier detection of market trends
A double-digit MAPE improvement in forecast accuracy.
Enough time to adjust production before a turn: savings when the market went down, and demand met when it went up.
A structured process across all markets, with significantly stronger management support.
How it was done
Seasonal patterns
The data was cleaned and the seasonal pattern identified and removed before modelling, then added back to the forecast.
Benchmark forecasts
Univariate models built on historical sales alone set the benchmark that the more advanced models had to beat.
Specific leading indicators
A VAR model with a Lasso penalty tested the possible combinations of indicators and kept the group that best predicts future sales.
Multivariate forecasts
A library of multivariate models forecast the main variable together with the indicators identified.
Weighted by accuracy
Some models are better at short, mid or long horizons. Indicio combines them into one forecast, giving more weight to the best model at each step.
Figures in the original case study do not represent actual company data, for confidentiality reasons.