Forecasting, or the ability to predict important business figures, from production volume to turnover, creates a significant competitive asset, however, 2020 data may be distorted.
Although the pandemic data is not inaccurate or incorrect, it is unusual because it is unlikely to give any indication of future trends.
Forecasters should beware of data drift, which is essentially an unusual term for GIGO: Garbage In, Garbage Out.
When this data is fed into previously accurate forecasting models, GIGO takes control and produces inaccurate results.
The obvious solution is to repeat prediction algorithms and embed real-time machine learning or other technical improvements in your prediction processes that are less based on historical data.
This can be a costly and time-consuming effort, especially for companies that do not have significant internal data science.
Identify key processes, and the people who interact with them, where predictions such as sales and marketing, production planning, staffing, or forward-looking areas can play an important role, are good places to start your research.
The ultimate goal is to identify the people who use forecasting data to do their job and influence their decision-making.
Make sure that consumers of forecasting data and their management are aware of the risks associated with the use of forecasting tools.
The results of these tools, which do not pass the snooping test, should be investigated and questioned.
It is very encouraging for entrepreneurs to form a small team of data and forecasting experts who can help their colleagues revise forecasts, or run ad hoc runs that ignore anomalous data.
In too many companies it is forbidden to “questioning the machine,” whether explicitly or unofficially.
By enabling employees to use their own experience and knowledge, we ensure that the company is not misled by forecasts fed with data from extraordinarily odd years.
For more information, read the original story in TechRepublic.
