In this video, Camil Blanchet, Ikigai's Enterprise Solution Architect, answers one of the more frequent questions that we get from our clients: which data inputs to use for demand forecasting.
Watch VideoOne question we often get is which data stream should be included in a demand forecasting process. Now there is really a two-step iterative process you can go through to determine what external drivers you should include in your forecast.
The way it works is first you start wide: you include everything you can: weather, macroeconomic, or industry-specific metrics, consumer sentiment based on social media, or anything else that you think could be a leading indicator and help boost your forecast accuracy.
Next what you do is you prune down those external drivers based on forecast value added so you make your forecast based or incorporating the data streams that I just mentioned or that you choose to include and then over the fork at the end of the forecasting period that's relevant to your forecast you look back and calculate the forecast value added from each step in your forecasting process. This helps you prune down which external drivers are not contributing to forecast accuracy and highlights the value added to including certain external drivers.
In this video, Camil Blanchet, Ikigai's Enterprise Solution Architect, answers one of the more frequent questions that we get from our clients: which data inputs to use for demand forecasting.
One question we often get is which data stream should be included in a demand forecasting process. Now there is really a two-step iterative process you can go through to determine what external drivers you should include in your forecast.
The way it works is first you start wide: you include everything you can: weather, macroeconomic, or industry-specific metrics, consumer sentiment based on social media, or anything else that you think could be a leading indicator and help boost your forecast accuracy.
Next what you do is you prune down those external drivers based on forecast value added so you make your forecast based or incorporating the data streams that I just mentioned or that you choose to include and then over the fork at the end of the forecasting period that's relevant to your forecast you look back and calculate the forecast value added from each step in your forecasting process. This helps you prune down which external drivers are not contributing to forecast accuracy and highlights the value added to including certain external drivers.
