EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
1350. Not Just Small Samples: Successfully Predicting Revenues With the Ratio Heuristic
Invited abstract in session WD-11: Heuristics in BOR, stream Behavioural OR.
Wednesday, 14:30-16:00Room: 12 (building: 116)
Authors (first author is the speaker)
| 1. | Florian Artinger
|
| Business Administration, Berlin International University of Applied Sciences |
Abstract
A key psychological principle underlying intuitive forecasting is the use of ratios. A finding is that heuristics, which operationalize such principles, can only predict well in comparison to more complex models when the training sample is small. Challenging this finding, we build on the bias–variance trade-off to develop three conditions under which a simple heuristic can match or even outperform more complex prediction models. To test these conditions empirically, we investigate revenue predictions using the “ratio heuristic”: multiply the quantity—revenue, in our case—observed in the first t time units by a constant to predict the future quantity. On 20 prediction tasks, the results are in line with the hypotheses showing that a smaller sample, a longer observation period, or unpredictable changes over time provide performance advantages for the heuristic such that it can outperform more complex, standard prediction models. Moreover, given sufficiently strong unpredictable changes over time can result in a setting where even a large sample and a short observation period do not provide performance advantages for the more complex forecasting methods. That is, managerial heuristics may hold value for business forecasting even in data-rich settings.
Keywords
- Forecasting
- Decision Analysis
- Behavioural OR
Status: accepted
Back to the list of papers