Advanced · Data and analysis · module 139 of 155
Forecasts and uncertainty
Use AI for a forecast, and make the uncertainty visible instead of promising one precise number.
After this lesson
You recognise the limitations of a forecast, calculate with a range and communicate uncertainty honestly.
In view · Choices and control
Forecasting with a range
The same period in previous years; look at exceptions separately.
A cautious and an optimistic estimate, with the assumptions included.
Prepare for both outcomes and compare with reality afterwards.
Part 1 of 3 · Compare fairly
Your preference for less movement is on. Choose each part yourself.
Extending a trend is risky
Three months of growth does not mean the growth will continue like that. Seasons, one-off events and chance play a part. So compare with the same period in previous years, not just with the month before. AI can make a simple forecast or suggest a formula (lessons 78 and 109), but the model does not know your environment: a new competitor, a renovation or a change in the law is not in your figures. Note such factors yourself.
Work with a range
A forecast of exactly 52.37 registrations suggests an accuracy that is not there. Give a range instead, for example between 40 and 60, with the assumption it is based on. Also look at a cautious and an optimistic scenario. The less data you have, the wider the range must be. Ask AI to list the assumptions separately and check whether they match what you know. One exceptional week or month can heavily distort an average.
Plan with the uncertainty
Use a forecast to be prepared, not as a certainty. For example, plan the room and materials for 50 participants, with an agreement on what you will do at 40 or at 60. Afterwards, compare the forecast with reality and note what you learned (lesson 67). Communicate the uncertainty to whoever decides: an honest range helps more than a precise number that does not come true (lesson 129).
New practice example
From explanation to a usable result.
A forecast is an estimate with uncertainty. Work with a range and scenarios, and state your assumptions.
Fictional practice situationFictional: last year, a youth centre counted an average of 35 visitors per Friday evening. One Friday with a free concert drew 140 visitors.
View a possible solution
This is how it can be done
The average including the concert evening gives a distorted picture. I also calculate the average without that exceptional evening and forecast 30 to 40 visitors for an ordinary Friday. For an evening with a free concert, I make a separate estimate.
Why this works
One exception pulls an average up sharply. By looking at it separately, the forecast for ordinary evenings becomes usable.
Make it slightly more difficult
Which event that is not in your figures could have the strongest influence on your next forecast?
Extra practice, at your own pace. You do not need to submit anything. This does not count towards an automatic assessment.
Your workplace · learning by doing
Make something. See what changes.
Checking the AI connection… You can already view the written example.
Present your assignment to Ende.
Read the answer, change one thing, and compare.
Previously held
New version
You remain in control
What do you want to improve in your next attempt?
On this device only. AI answers remain only in this page memory. This is self-review, not a grade.
Why this approach works
One exception pulls an average up sharply. By looking at it separately, the forecast for ordinary evenings becomes usable.
A step further
Which event that is not in your figures could have the strongest influence on your next forecast?
Try & discover · no obligation
Become the source checker yourself
Only mark statements that the source does not support. Then check your selection.
Fictional registrations for summer camps: 41, 48, 46 and 55 over the past four years. A colleague discusses the forecast for next year.
Select which sentences you would correct above. The answer remains hidden until you check your selection.
This trial works locally with fictitious data and does not count towards your lesson completion. It starts anew with a new lesson or reload.
A prompt to get you started
Here are the fictional registrations for our summer camps: 2022: 41, 2023: 48, 2024: 46, 2025: 55. Make a forecast for 2026 as a range, with a cautious and an optimistic scenario. List the assumptions separately and say what information is missing. Do not give a number with decimals.
Apply it
After three months of rising sales, a shop has AI forecast the whole year. The forecast shows a doubling, and the owner wants to hire extra staff straight away.
Which approach best fits this situation?
- Hire straight away on the basis of the doubling.
- Compare with the same months in previous years, create a range with scenarios and first make a decision you can adjust.
- Ask a second AI tool to confirm the same forecast.
With JavaScript, you can check your approach and receive immediate feedback.
Test your insight
An AI tool forecasts “exactly 52.37 registrations” based on four years of figures. How do you use that?
- Provide exactly 52 places.
- See it as a rough estimate, make a range and plan for several scenarios.
- Ignore the forecast, because forecasts are never right.
Extra practice
Choose a figure from your work that you want to forecast. Make a range with two scenarios and write down what you will do if the result comes out at the low end or the high end.
Forecast for 2026: between 45 and 60 registrations. Cautious scenario, 45: the rise in 2025 was a one-off, due to a new flyer. Optimistic scenario, 60: the growth continues. Missing: the number of places at other camps in the region. Plan: materials for 55, a waiting list from 56 and an extra supervisor if there are more than 50 registrations.
