The situation
A tier-1 supplier in the chemical industry, for which the automotive industry is one of the main markets.
It needed to predict demand in its markets, with insight into sales volumes and their drivers, to guide market strategy.
The challenge
- Little clarity over market shiftsDemand could not be forecast accurately, so the company could not respond to fluctuations or predict supply on time. Without visibility of macroeconomic effects, resource allocation led to unnecessary costs and lost revenue.
- No leading indicatorsA bottom-up approach limited forecasting to simple univariate models, which cannot use leading indicators, even though the company had enough data for multivariate models.
What Indicio did
- Identified the leading indicators that best forecast demand for its chemical products.
- Built forecasts for each product group and region, updated quickly every month.
- Replaced a process without forecast models with 50+ econometric models, weighted into one forecast.
The result
- 56%
- more accurate forecast: error (MAPE) of 7.96% against 11.69% internally
- 18–56%
- improvement at 2 to 11 months ahead
- 50+
- econometric models weighted into one forecast
A double-digit MAPE improvement, from 18% to 56% across 2 to 11 months ahead.
An overview of market drivers by industry and by region.
Visibility into how the relevance of each leading indicator changes over time.
With the models built, a quick monthly update is all it takes.
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.