Academy: Why use quantitative methods


Predicting the future sales is not a trivial task. The product port folio might be big, competitor impact our sales, we get a noisy signal because our customers are far from end consumer, and many more factors making forecast inaccurate.

Add to here the noise the company itself create: New product launched, campaigns/deals, reallocation of warehourse inventory, quarter-end sales push, frequent changes of forecasts, etc.

It is complex and time consuming for the human brain to filter all those factors into a realiable sales forecast. The quantitative forecast is able to do this task efficient, and even beat the pure judgemental forecast in 90% of the cases[1].

The quantitative forecast covers a set of different approached. Research have identified good and bad methods and new research are launched frequently. At sophub we acknowledge research and utilize latest findings, and combine it with our own methods and learning.

Using a statistical model should always be driven by selecting the most accurate model that fits the pattern for each single item in an out-of-sample approach.

In fact when sophub engage a new client we do it in the context of research. We prove we are better than the current used forecast and also set a target for how good a forecast you will be able to get by using sophub approach.

sophub uses the hybrid quantitative method called autonomous forecast to generate the most accurate forecast a machine can generate. The autonomous forecast utilize both statistical and artificial intelligence.

In the M4 research competition the hybrid method was recognized to be superior to other methods. sophub can confirm this on our client port folio[2].


[1] Byskov, Brian (2016), sophub research based on 15 companies

[2]Makridakis, Spyros (2018), The M4 Competition: Results,findings, conclusion and way forward, International Journal fo Forecasting, 34(20), 802-808