How do you predict? what kind of prediction algorithms/tools/methods/technologies you use?
Prediction? Please be more specific. What you want to predict? For example, you want to predict market performance? to predict the future? to predict football?, to predict what? ... all have different means. Anyway, It is a very risky business to be in, because no one has a crystal ball. It is rather general question.
Some of the heuristics I rely on:
(1) I think of the world as a tangle of stocks/accumulations and flows (inputs/outputs) into those stocks. Stocks provide stability, inertia, and memory, and hence constitute a basis for extrapolation into the future (they don't change over night: for example the fleet of vehicles on the streets has a full replacement time horizon of 10 to 13 years);
(2) I separate phenomena into convergent (trends, natural histories, causal structures with high necessity and low contingency) and divergent (highly contingent events) and realize that convergent phenomena are likely to continue alongside their path into the future whereas divergent phenomena are impossible to predict. At the logical intersection of the two sets we have the interesting class of "improbable inevitabilities" - particular actualizations of necessary phenomena, that could have actualized in a different manner (hence the improbability of those inevitabilities);
(3) Other things being equal, the near term is more directly influenced by the existing structure of stocks, and hence it is easier to predict than the longer term, and this should be reflected in one's confidence calibrations.
(4) Always useful to think in terms of several dynamic scenarios, trying to capture how the present would morph into a set of alternative futures;
(5) Always useful to separate as explicitly as possible what you think will happen from what you wish would happen and things within your sphere of influence from things beyond your sphere of influence.
(6) Forecasting involves distancing oneself from the present and we need to be aware of the mental representations involved in the process of subjective distancing.
Article Beware of contingency
Article Proximity, subjectivity, and space: Rethinking distance in h...
How to predict?
Without knowing in details what are the target to predict e.g. predicting the future events / happenings, predicting / forecast stock prices, independent variables predicting dependable variables in a research etc., high level general prediction can be done in the following steps:
Prediction? Please be more specific. What you want to predict? For example, you want to predict market performance? to predict the future? to predict football?, to predict what? ... all have different means. Anyway, It is a very risky business to be in, because no one has a crystal ball. It is rather general question.
Dear Amir,
Predictive analysis is highly risk analysis.But it has helped in assessing health risk , detecting fraud to go beyond the descriptive statistics.
In brief Predictive analytics is the use of data , statistical algorithms (PLS), machine learning,artificial intelligence techniques to identify the likelihood of future outcome based on historical data.
I agree with the insightful views of the colleagues.
'A rational prediction has an explanation based on theory.'
-W. Edwards Deming
Regards
Dear Amir
First, I would like to thank you for inviting to participate in an interesting but intriguing question.
We have different methods to predict. There are analytical methods if events of interests have a definite pattern or trend in time series, there are approximate methods in which events varying in space or over time have a pattern or trend not so definite to predict by analytical analysis, we can predict with some conviction by an approximate function, there are also statistical methods to predict. But, a word of caution as pronounced by Jayanta Kumar Maji ·as there is "the principle of uncertainty". We cannot be 100% sure that our prediction would come true. In fact, reality is a web of interrelations and everything is connected to everything else. Therefore, we cannot account for all variables that are involved in the occurrence of an event, We take into account all those variables or events which are known to cause an event or theoretically understood so.
By analysing the pattern in the data set, we can formulate some rules for prediction.
Long ago, some people were predicting the emergence of accidents or the birth of the prophets by astronomical calculations. Also, inspirational predict but be misleading.
A prediction is a forecast about future, but not only about the weather, may be regarding any aspect.. Pre means “before” and “diction” has to do with talking. So a prediction is a statement about the future. It's a guess, sometimes based on facts or evidence, but not always.
It is very important to know what type of object/ behaviour you want to predict. There are different methods, many available articles, as well as many specific questions about prediction, over 1000. I have attached links to them.
This is fine theoretical paper: A Prediction System Based on Fuzzy Logic
The main objective of the paper is to build a prediction system to predict the future occurrence of an event. Fuzzy logic, among the various available Artificial Intelligence techniques, emerges as an advantageous technique in predicting future events. Subjective and Objective modeling are two types of fuzzy modeling. Objective type fuzzy modeling is used to build the prediction system. It is a combination of a clustering algorithm and fuzzy system identification which proves effective in improving the efficiency of the prediction. To train the prediction system, historical data is obtained from the web. Data specific to the desired application is obtained and is recorded. This recorded information is subjectively reasoned to develop containing only the necessary inputs to the prediction system. The subtractive clustering algorithm is used for its computational advantages and fuzzy rules are formed using system identification technique...
https://www.researchgate.net/search.Search.html?query=prediction%20methods&type=publication
https://www.researchgate.net/search.Search.html?query=how%20to%20predict&type=question&tabViewId=581c460af7b67e40d2152a3e
http://www.iaeng.org/publication/WCECS2008/WCECS2008_pp804-809.pdf
It is very important the level of information (scales of measuremnt of the empirical information) for the choice of the prediction procedures.
Prediction or forecasting?
Anyway, prediction:
- find if there exists so, the underlying law that governs your system
- observe the time stamp where you are now
- apply proper techniques to estimate parameters
- use them to predict the solution for next times
That's a common route.
I used below methods for prediction using Clementine Software. These methods were decision trees from Data mining approach.
Wisdom based on knowledge and experience, as well as our professional and human intuition can help us to predict some situations in science and in life.
Today, in the 21st century, it is very important to use the knowledge acquired and new technologies not only to predict the negative events, but also to prevent (at least to try!) them and to reduce or avoid the damages. For example, if we are able to predict progression and outcome of the disease, it will help us to use the most adequate preventive and treatment strategies.
Dear all,
Prediction as difficult as it may be, is assigning a probability value on a presumed would be event. We only are certain what is happening now and what has already happened, not what will happen next to what is happening now. Events are points in space time and we extrapolate what we developed as functions or properties or paths of phenomena we studied or know with accumulated experience and knowledge to make predictions of events that lie on those developed paths, how zigzag or twisted they may be. Formulas give us exact future values of functions with probability value one but events do not be fully governed by formulas we developed for them as we do not know all parameters of nature in an interval of time, the events are under.
Dear Amir,
In my opinion, the most initial step is interpreting the phenomenon of 'prediction'. Relying on phenomenographic interpretation of prediction, we can come up with two main issues: (i) quantitative prediction and (ii) qualitative prediction. The first item is highly dependent on 'quantitative measures'. By taking quantitative measures into account we will be concerned with different kinds of 'predictive mappings' (from a phenomenon into another phenomenon). The second item is concerned with empirical and experience-based approaches and methodologies. However, it shall be stressed that both items have their own mathematical, statistical, computational and logical dimensions, and thus, we will need to design mathematical, statistical, computational and logical models for simulating our problems in both quantitative and qualitative predictions. The second similarity between (i) and (ii) is their special focus on discovering (and even suggesting) different 'patterns'. Patterns are capable of discovering different hidden interrelationships between multiple phenomena in the world.
Kind regards,
Farshad
The prediction of the future comes due to some scientific facts, for example, water scarcity, rise of sea level, changing the balance of powers in world… etc.
Prediction of the future should be started with the question of type of development. In case of development which follows the "negative feedback loop" idea the prediction of the future could be done by using exact mathematical expressions. In case of "positive feedback loop" then the future could be estimated by using e.g. the "density matrix" conception. But instead of focusing of the discussion about the future state, the discussion should be enlarged and considering also the consequences for the entity, because this is the aspect which follows the imagination of the future. Working on a cybernetic approach for the management which considers both developments.
We may use some statistical methods for forecast or prediction. This depends, of course, on previous data. Regression analysis is one of the best methods used for prediction.
This is one field of prediction application.
Mix monitoring tools to measure machine health! What is your equipment trying to tell you? Yes, this is about predictive maintenance.
Predictive maintenance is synonymous with reliability as it’s one of the tools reliability engineers depend on to create cost savings that give plants a competitive edge.
Predictive maintenance technicians use many technologies that can provide plants with the information necessary to identify faults. When used as intended, this can and will provide process, quality, and equipment reliability...
People can use past 'patterns' to 'predict' future things. For example, your co-worker usually leave office at 12-1 PM for lunch. Today, you don't see him/her around this time, you can predict that he is most likely out for lunch. Another example, a university always ranks among top 3 around the world for the past 20 years, then we can predict that it should be in this range of rank in the next year (with very high chance).
Prediction can be made by numerical methods. This is the case of meteorological previsions.
Want to predict how someone will vote? Identity may matter more than policy!
Researchers say existing models for predicting voting behavior focus too much on policy positions, and not enough on voter identity...
Voting as identity reinforcement has an immediate effect. You check the box or tap the screen and feel that you’ve acted in accordance with your self-image. In contrast, one vote rarely sways an election, and the effects of an election on policy are often not seen until far in the future, after a candidate has taken office, if at all.
Despite this, traditional voter prediction models still put questions about policy front-and-center. Voters are asked to rank policy issues by importance, and the model selects the candidate whose platform best matches their positions. It’s a familiar format to anyone who’s used an online tool to calculate which candidates most agree with you on election issues. “The models assume that people make decisions as if they were considering the costs and benefits of each relevant factor, weighted by the importance of that factor,” Huettel says...
https://www.researchgate.net/blog/post/want-to-predict-how-someone-will-vote-identity-may-matter-more-than-policy
Equipped with appropriate data you can estimate (predict??), quite precisely, the number of people killed by lightning stroke next summer. But not their names. Not everything is predictable, so there is no single answer the question posed here. In simplest cases the obvious approach is extrapolation of existing (or smoothed) data already known.
Dear Amir Mosavi,
Prediction techniques are different in individual fields of knowledge. In economics and finance, quantitative instrumental methods based on the analysis of quantitative data and statistical analysis are used. In addition, extrapolations of the trends of diagnosed processes that have been measured and tossed over a specific period of time in the past are used. while in the humanities, in knowledge fields characterized by qualitative specificity, other forecasting methods that may be partly qualitative are also used. In biological sciences, methods for observing changes in various natural, climatic, cosmic phenomena, etc. are used.
I wish you the best in New Year 2019. Best wishes
- Badges
- Science topic