The prognosis ( ancient Greek πρόγνωσις prognosis 'prior knowledge' or 'advance knowledge'), German prediction or prediction , rarely also: prediction ( Latin praedicere ' to predict') is a statement about events , environmental conditions or developments in the future . The prediction has a different time course than the retrodiction and explanation . Forecasts differ from other statements about the future (e.g. prophecies) in their scientific orientation.
A valid prognosis is based on facts, which are often collected using formalized methods ( measurements , time-structured series of measurements or simulations ) to create data . On this basis, predictions can then be made and decisions made with a certain probability . The data on which the forecast is based are referred to as (better or worse) predictors . In contrast to pure intuition include begründbares experiential knowledge and its extrapolation to the accepted forecasting methods. Such argumentierbaren predictions are methodologically important in all areas of science.
- An essential feature of the decisions in each area is their future orientation:
- Decisions are always based on forecasts or prognostic expectations.
- Decisions have to be made objectively under uncertainty . You are at risk because the decision-makers only have imperfect information.
- There is an additional difficulty in the area of sociological prognoses: the “objects” of the prediction are themselves actors (“ subjects ”) and could change their behavior based on the prognosis. ( See the "self-destructive" and the " self-fulfilling prophecy ".)
The epistemological consideration of prognoses is closely connected with the concepts of causality and predictability , and in the implementation also with fundamental aspects of probability and chance . In empirical research, the prognostic validity is an important quality criterion for the operationalization of constructs .
- The science of general prediction is futurology .
- In physical measurement technology, one speaks of the expected value .
A number of the prerequisites for an accurate forecast are identified, including:
- Not triviality : The following statement pattern should not occur: "Tomorrow it will rain or not."
- Objectivity : Verifiability of the method, this also includes the complete specification and specification of the conditions (so-called framework conditions ) on which the arrival of the predicted result is made dependent.
- Validity : Is what is actually forecasted what should be forecasted?
Types of forecasting techniques
A prognosis method should be better than the naive prognosis , otherwise the additional effort compared to the naive prognosis is not worthwhile.
Forecasting techniques can be classified in different ways. With regard to their horizon, short, medium and long-term forecasts can be distinguished. In addition, a distinction is made between qualitative and quantitative techniques. In addition, they can be divided into “top-down” and “bottom-up” with regard to their creation perspective. The simplest forecasting method is the naive forecast.
Qualitative forecasting techniques
- are subjective assessments that are created intuitively by experts with mature specialist knowledge
- One possible variant is linear extrapolation → past values are roughly projected into the future
- further variants → opinion polls or life cycle analyzes
- try to anticipate trends
- are more complex
- rather provide few concrete figures
Quantitative forecasting techniques
- consist mainly of the processing of data material
- give concrete, numerical results
Top-down or bottom-up forecast
The top-down forecast approach is centralistic and is particularly suitable for stable demand situations. For example, if a company has four distribution centers whose demand was 4: 3: 2: 1 in the past, an aggregated demand quantity based on the demand of the entire market would be distributed to the distribution centers in a corresponding ratio.
With the bottom-up forecasting method, each distribution center would create its own forecasts and transmit them to the production facility, where they would be aggregated. The method takes into account regional market developments, but is more difficult to organize.
Despite all efforts to technically optimize forecasts, there will always be larger or smaller deviations between the forecast and the event that actually occurs. It is therefore very important - also when choosing the correct forecast model - to evaluate the quality of the selected or the considered method by determining the forecast errors.
In the context of qualitative forecasting , forecast errors cannot be quantified in advance. The causes of errors include a .:
- Younger values are overrated.
- Values that are currently popular or much discussed are overrated.
- Apparent patterns are recognized, but they do not exist empirically.
- Special events are remembered, while normal ones are quickly forgotten.
- Desires or fears can flow into prognoses.
- Propensity to select, search for and / or interpret information in such a way that it meets one's expectations (see confirmation errors ).
With quantitative forecasting , the forecast accuracy is evaluated using the forecast error that is determined. The most common procedures are briefly listed below:
- Root mean square deviation ( english mean squared error , in short: MSE )
- Mean absolute deviation from the median (English Median absolute deviation , MAD for short)
- Mean absolute percentage deviation (English Mean Absolute Percentage Deviation , abbreviated: MAPD )
MAPD indicates a relative value, which means that it offers other possibilities for comparison than the MSE and MAD, which are given in absolute numbers.
Examples of important quantitative forecasting techniques
- One-dimensional procedure
require a large amount of data, they provide poorer values for long-term forecasts and often provide poor forecasts even with strong sales fluctuations. However, they can be systematized well and used for a large number of products. In addition, they are easy to understand. Well-known one-dimensional methods are: exponential smoothing, trend forecasting, moving averages ; Here are rolling averages used.
- Multidimensional processes
are based on the causality of the sales figures for various variables, such as price and promotions. It is assumed that the sales are influenced by factors such as B. the weather with ice or the season of mineral water is directly related. Well-known multidimensional methods are: econometric models and regression analysis .
Procedure in detail:
- Basic mathematical and statistical methods are extrapolation and projection .
- Trend forecast : projection of a series of values into the future.
- Exponential Smoothing : Exponential smoothing is a forecasting method that predicts future values based on past values.
- Regression calculation : Analysis of functional relationships between at least two quantities.
- Econometric models: Analysis of economic relationships based on the formation of overall models with many variables and statements about the relationship between all these variables.
- Portfolio analysis : mostly graphically oriented analysis of two or sometimes three sizes.
- Life cycle analysis : Analysis of the course of a development over time. Requires precise market observation and market research.
Examples of important qualitative forecasting techniques
- Delphi method : This is a written, multi-phase survey of experts, with each new round of questions informing them of the results of the previous round. Some of the respondents are asked to justify their answer. These reasons are used by all respondents in the next round to review their opinions and to change them if necessary.
- Scenario technique : Representation of several hypothetical sequences of events in order to identify causal relationships and decision-relevant milestones.
- Relevance tree analysis : Isclose to game theory . Retrograde derivation of possible solutions for given situations based on decision theory.
- Historical analogy : analysis of a development over time. Market-specific details are taken into account to a large extent.
Areas of application in the field of scientific modeling
- Weather forecast ( synoptic meteorology ), climate models
- Forecast of floods
- Forecast of volcanic eruptions
- Forecast of disease or pest infestation in agriculture
Areas of application in the field of human sciences
Politics and Political Science
In addition to the political parties 'and politicians' think tanks, which are not always public in their choice of methods, there are special commissions for forecasting purposes - for example, the German Federal Government has the Advisory Council for assessing macroeconomic developments or the Protection Commission at the Federal Ministry of the Interior . The governments themselves also submit forecasts, such as their annual economic reports and the annual budget , which in the cameralistics shows the forecast as planned. International organizations such as the OECD , IMF and the EU Commission also have corresponding sub-organizations or advisory boards and issue forecasts.
In demography , prognoses in the form of population prognoses based on assumptions about the future development of fertility , mortality and migration play an important role. The Federal Statistical Office carries out such forecasts for Germany .
In business administration, the prognosis is often referred to as a forecast . Various both qualitative and quantitative forecasting methods can be used in many areas of application (selection):
- Long-term forecast of sales opportunities and market potentials for new products as part of production planning and control ( Delphi method ).
- Sales forecast of products with a large number of variants (e.g. vehicles) using a tiered process model with a hybrid approach
- Sales forecast of a product with special consideration of the growth potential of specific sub-markets
- Deriving sub-goals and strategies, for example for developing long-term strategies (relevance tree method).
- Prediction of product life cycles for new products (historical analogy).
- Inventory forecast ( needs assessment , trend forecast, exponential smoothing).
- Sales forecast under stable conditions (trend forecast, exponential smoothing).
- Prediction of uncontrollable production and loads in power grids
- Forecast of price centers and inflation
Economic forecasts are usually prepared in spring and autumn for the current and the coming year. Medium-term forecasts include several more years to come. Long-term forecasts are based on decades. Most macroeconomic forecasting institutions are under public law, some companies - such as the big banks - also have their own macroeconomic departments that create macroeconomic forecasts.
- Please refer:
- Joint forecast of the economic research institutes
- Economic forecast
- Another area is the prognosis of price developments ( stock exchange prices , exchange rates )
When it comes to economic decision-making processes, one speaks of controlling .
Medicine, dentistry and veterinary medicine
In medicine , the term prognosis has been used to describe the assessment of the course of the disease since ancient times , with mantic elements also serving for prognosis. If the probability of a cure is high, the prognosis is good; if it is low, the prognosis is bad. If there is no short to medium-term survival probability, the term poor prognosis is used.
Treatment can change the prognosis as the disease progresses. It depends on the diagnostics and treatment options available . Concomitant illnesses , compliance and social factors such as education and financial situation also play a role.
One of the prognostic signs is, for example, the hippocratic face of patients who are expected to die soon.
“The terms prognosis and prediction are clearly differentiated in oncology .
- Prognosis describes the statistical probability of a (breast cancer) recurrence, a locoregional recurrence or distant metastasis, or a death caused by breast cancer.
- Prediction means the relative prediction of an effect based on therapeutic intervention (e.g. primary systemic therapy, adjuvant therapy, surgery, etc.).
Both definitions are not based on individual data but describe statistical probabilities; consequently, individual predictions beyond the uncertainty of relativizing influencing factors are prohibited. "
In veterinary medicine , a distinction is made between a prognosis quo ad vitam and a prognosis quo ad usum for farm animals . The prognosis quo ad vitam describes the chance that the animal will survive the disease, quo ad usum the chance that the animal can be used again as a farm animal (riding horse, dairy cow, carrier pigeon etc.) after healing.
In the legal assessment of issues, prognostic decisions of various kinds can be necessary. However, they are of independent importance, especially in the area of criminal law , because a number of decisions, in particular the determination of sentences and the execution of sentences, must be made on the basis of a prognosis of the offender's future behavior. It is also known there under the term social prognosis or crime prognosis .
The history of religion offers a multitude of comparative cases, the sociology of religion current data. This is the foundation on which approaches to prognostic religious studies begin - about the consequences of discrimination, about the development of institutions or about the development of fundamentalist currents, etc.
- A prediction based on faith and divine calling is a prophecy or prophecy, in the narrower religious sense a prophecy .
- In the field of esotericism and astrology (see. Also history of divination ) is called divination , soothsaying, precognition or clairvoyance .
Like all social sciences , sociology also has the problem that its prognoses can be heard by the objects of their prognosis, which can then follow it accordingly or counteract it. To this end, see self-fulfilling prophecy and Epignose .
The Quantitative Linguistics provides opportunities available in the field of language change processes under certain conditions to forecast. The prerequisite for this is that linguistic changes are recorded quantitatively over a longer period of time; The number of borrowings of words from other languages into German ( loan and foreign words ) is partly quite well recorded. Since it is known that these processes usually proceed according to the so-called Piotrowski law , one can dare to make predictions for most of these developments, at least for the near future, without taking too great a risk of misjudgment. This has been shown by computer experiments in which forecasts based on past centuries were simulated in the statistically recorded present so that a control over the quality of the forecast was possible. In the case of borrowings from Latin and English into German, it turned out that a prognosis about the further development of the Anglicisms is less certain than a prognosis about that of the Latinisms . The two processes differ in that in the case of the Latinisms the turning point of the development can be determined with some certainty, in the case of the Anglicisms not yet.
The opposite perspective is also possible: the retrospective or retrodiction . If the later development of a language change has been recorded statistically, while the early one cannot be observed, this early course can be inferred with the help of Piotrowski's law. Kohlhase was able to quantitatively record the gradual transition from “ward” to “became” for the 1st and 3rd person singular indicative past tense of the verb “werden” from 1467 onwards with the Nuremberg chronicler Heinrich Deichsler and based on this data statements about the beginnings of this change in win its idiolect .
Controversial use of forecasts in the formation of political opinion
Critics complain that predictions are often used to influence individual behavior or public opinion. They are particularly to be questioned critically if they make statements over long periods of time or in dynamic systems or if they are in the self-interest of the forecasters. The criticism of forecasts is expressed in a variety of ways in the context of Internet forums, reports, non-fiction programs or political cabaret. Criticized topics include:
- Forecasts for population and pension developments
- medical prognoses for health developments in diseases such as diabetes and obesity
- Share performance forecasts
On a factual level, the criticism is often based on the fact that forecasts can represent future changes using only the data of the past and the theories of the present. Forecasts aim to predict what the future will be like. In the present, they can only be criticized with regard to their premises and data basis.
- "Predictions are difficult, especially when they concern the future." (Attributed to Karl Valentin , Mark Twain , Winston Churchill , Niels Bohr , Kurt Tucholsky and others)
- "The best way to predict the future is to invent the future". ( Alan Kay , computer scientist)
- "A forecaster is a man who has somber premonitions in clear moments". ( Tennessee Williams )
- Peter Mertens , Susanne Rässler (Ed.): Forecast calculation . 6th edition. Physica, Heidelberg 2005, ISBN 3-7908-0216-6 .
- Overview of forecasts from 1910 on the media landscape in 2010 . From the book "The World in 100 Years", edited by Arthur Brehmer (1910).
- Herlyn, W .: PPS im Automobilbau - Production program planning and control of vehicles and aggregates, p. 147 ff. Hanser Verlag, Munich 2012, ISBN 978-3-446-41370-2 .
- Oliver Handel: Demand Endogenization of Intermediate Products in Supply Chains through a System-Dynamics-based Modularization Concept. In: System Dynamics Review, Delft, 2013
- Dirk Ulrich Gilbert, Vera Magin, Michael Müller: The challenge of price prognosis: What is the price of tomorrow. In: Marketing Review St. Gallen, 30, 5, 2013, pp. 98-109. doi : 10.1365 / s11621-013-0281-3 .
- John E. Hanke, Dean W. Wichern: Business forecasting . 9th edition. Pearson / Prentice Hall, Upper Saddle River, NJ 2009, ISBN 978-0-13-230120-6 .
- Donald J. Bowersox, David J. Closs: Logistical management: the integrated supply chain process (= McGraw-Hill series in marketing ). McGraw-Hill, New York 1996, ISBN 0-07-006883-6 .
- Thomas Schneckenburger: Forecasts and segmentation in the supply chain: a procedural model to reduce uncertainty . 2000, DNB 960278834 (dissertation University of St. Gallen).
- Horst Tempelmeier : Material logistics - models and algorithms for production planning and control in advanced planning systems . 6th edition. Springer, Berlin 2006, ISBN 3-540-28425-7 .
- Péter Horváth: Controlling . 12th edition. Vahlen, Munich 2011, ISBN 978-3-8006-3878-9 .
- Joachim Telle : Finds on empirical-mantic prognosis in medical prose of the late Middle Ages. In: Sudhoff's archive . Vol. 52, No. 2, 1968, pp. 130-141, JSTOR 20775660 .
- Michael Stolberg : The history of palliative medicine. Medical care for the dying from 1500 until today. Mabuse-Verlag, Frankfurt am Main 2011, ISBN 978-3-940529-79-4 , pp. 65-67.
- AGO manual ( ISBN 978-3-9501446-3-5 ) or archived copy ( memento of the original from October 25, 2011 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice.
- Lars Clausen , On the asymmetry of prognosis and epignosis in the social sciences. In: Ders .: Krasser Sozialer Wandel , Opladen 1994, ISBN 3-8100-1141-X .
- Helle Körner: On the development of the German (loan) vocabulary. In: Glottometrics 7, 2004, 25–49 (PDF full text ); Katharina Ternes: Developments in German vocabulary. In: Glottometrics 21, 2011, pp. 25–53 (PDF full text ).
- Karl-Heinz Best : Are prognoses possible in linguistics? In: Tilo Weber, Gerd Antos (Hrsg.): Types of knowledge. Conceptual differentiation and characteristics in the practice of knowledge transfer (= transfer sciences. Vol. 7). Peter Lang, Frankfurt am Main a. a. 2009, ISBN 978-3-631-57109-5 , pp. 164-175.
- Jörg Kohlhase: The development of ward zu was with the Nuremberg chronicler Heinrich Deichsler . In: Karl-Heinz Best, Jörg Kohlhase (Ed.): Exact language change research . edition herodot, Göttingen 1983, pp. 103-106. ISBN 3-88694-024-1 .