Naive prognosis

The naive prognosis is probably the simplest of all prognoses . It takes what is currently known and predicts that it will be the same in the future.

Types

The specialist literature distinguishes between two types. You use various measures as a basis for your forecast:

Type 1 ( no-change forecast ): the current value
Type 2 ( same-change forecast ): the current trend , i.e. the distance between the current value and the previous one

The following formulas result for the naive prognosis:

${\ displaystyle {\ mathsf {f_ {t-1}}} = {\ mathit {penultimate value}}}$
${\ displaystyle {\ mathsf {f_ {t}}} = {\ mathit {current value = last value}}}$
${\ displaystyle {\ mathsf {f_ {t + 1}}} = {\ mathit {forecast value}}}$

Type 1: ${\ displaystyle {\ mathsf {f_ {t + 1}}} = {\ mathsf {f_ {t}}}}$
Type 2: ${\ displaystyle {\ mathsf {f_ {t + 1}}} = {\ mathsf {f_ {t} + ({f_ {t}} - {f_ {t-1}})}}}$

use

The naive forecast is mainly used to assess the quality of another, more complex forecast calculation. For this purpose, the mean error of the prognosis to be assessed and the mean error of the naive prognosis are calculated and the quotient is formed from this. If this characteristic value is less than 1, the examined forecast calculation is better than the naive forecast or extrapolation . This means that the additional effort required for the more complex forecast was worth it.

Such an assessment is only possible after the actual occurrence of the event, since the actual value is required for the calculation.

Example of a naive forecast: number of power outages in a city, data shortly before the turn of the year

relative year	recorded power outages	Type 1 prognosis (same value)	Type 2 prognosis (same trend)
two years ago	1	no value	no value
Previous year	4th	1	no value
this year, so far	0	4th	7th
next year	no value yet	0	0 (if negative values were possible: -4)

literature

Horst Rinne: Pocket book of statistics . 4th edition German, Frankfurt / M. 2008, ISBN 978-3-8171-1827-4 .