Difference between revisions of "Forecasting"

From Glitchdata
Jump to navigation Jump to search
(Created page with 'Predictions of events based on data. There are several ways of forecasting. * Bayesian * ARMA - Auto Regression Moving Average * Technical Analysis - Finance ==Examples== * htt...')
 
(Applications of forecasting)
 
(44 intermediate revisions by the same user not shown)
Line 1: Line 1:
Predictions of events based on data.
+
Forecasting is the process of making statements about events whose actual outcomes (typically) have not yet been observed. A commonplace example might be estimation of some variable of interest at some specified future date.
  
There are several ways of forecasting.
+
Predictions of outcomes based on data and prediction models. Forecasting involves [[Predictive Modelling]].
* Bayesian
+
 
* ARMA - Auto Regression Moving Average
+
There are millions of datasets out there. There are thousands of models. They determine the quality of forecasts. Reducing error is always an issue.
 +
 
 +
* [[Forecasting History]]
 +
* [[Forecasting Application]]
 +
* [[Forecasting Methods]]
 +
 
 +
 
 +
 
 +
There are various forecasting methods listed below:
 +
==Extrapolative Methods==
 +
Extrapolative methods are to infer or estimate by extending or projecting known information.
 +
{|class="wikitable sortable" width=100%
 +
!#!!Method!!|Description!!Actuary Usage!! Business Usage
 +
|-
 +
| width=5%| 1
 +
| width=25%| [[Moving Average]], [[Simple Moving Average]]
 +
| width=40%|
 +
* [[Linear Regression]]
 +
| width=15%| Frequently used
 +
| width=15%| Widely used
 +
 
 +
|-
 +
|| 2
 +
|| Exponential Smoothing
 +
||
 +
* Holt-Winters methods
 +
|| Generally not used
 +
|| Widely used for TS analysis
 +
 
 +
|-
 +
|| 3
 +
|| [[Auto Regression Moving Average]] (ARMA)
 +
||
 +
* http://en.wikipedia.org/wiki/Box%E2%80%93Jenkins
 +
|| Generally not used
 +
|| Widely used
 +
 
 +
|-class="sortbottom"
 +
|}
 +
 
 +
==Explanatory Variable Methods==
 +
Use of a variable that explains a related response variable. Often called an independent variable or predictor variable.
 +
 
 +
{|class="wikitable sortable" width=100%
 +
!#!!Method!!|Description!!Actuary Usage!! Business Usage
 +
|-
 +
| width=5%| 1
 +
| width=25%| [[Regression Analysis]]
 +
| width=40%|
 
* Technical Analysis - Finance
 
* Technical Analysis - Finance
 +
| width=15%| Widely used
 +
| width=15%| Widely used
 +
 +
|-
 +
|| 2
 +
|| Predictive Modeling
 +
||
 +
|| Gaining popularity
 +
|| Widely used
 +
 +
|-
 +
|| 3
 +
|| Artificial Neural Networks
 +
||
 +
* http://www.neural-forecasting.com/neural_forecasting_data.htm
 +
|| Generally not used
 +
|| Sometimes used
 +
 +
|-
 +
|| 4
 +
|| Econometric Modeling
 +
||
 +
|| Generally not used
 +
|| Widely used
 +
 +
|-class="sortbottom"
 +
|}
 +
 +
==Simulation Modeling Methods==
 +
{|class="wikitable sortable" width=100%
 +
!#!!Method!!|Description!!Actuary Usage!! Business Usage
 +
|-
 +
| width=5%| 1
 +
| width=25%| Cell-based Modeling
 +
| width=40%|
 +
| width=15%| Frequently used
 +
| width=15%| Frequently used
 +
 +
|-
 +
|| 2
 +
|| Systems Dynamics Simulation
 +
||
 +
|| Generally not used
 +
|| Becoming more widely used
 +
 +
|-
 +
|| 3
 +
|| Multi-agent Simulation
 +
||
 +
|| Generally not used
 +
|| Becoming more widely used
 +
 +
|-class="sortbottom"
 +
|}
 +
 +
 +
 +
==Judgmental Methods==
 +
* Composite forecasts
 +
* Surveys
 +
* [[Delphi Method]]
 +
* Scenario building
 +
* Technology forecasting
 +
* Forecast by analogy
 +
 +
==Composite Methods==
 +
{|class="wikitable sortable" width=100%
 +
!#!!Method!!|Description!!Actuary Usage!! Business Usage
 +
|-
 +
| width=5%| 1
 +
| width=25%| Bayesian Forecasting
 +
| width=40%|
 +
| width=15%| Generally not used
 +
| width=15%| Generally not used
 +
 +
|-
 +
|| 2
 +
|| Others?
 +
||
 +
|| Generally not used
 +
|| Generally not used
 +
 +
|-class="sortbottom"
 +
|}
 +
 +
 +
==Forecasting==
 +
* [[Predictive Analytics]]
 +
* [[Machine Learning]]
 +
* [[Backward Propagation]]
 +
 +
==People==
 +
* Rob Hyndman - http://robjhyndman.com/
 +
 +
 +
 +
==Forecasting Models==
 +
Every industry has its "models". These range in the thousands.
 +
 +
* 5-20day Moving Average - Trading
 +
* Rainfall-Runoff Models - Hydrology
 +
* Logistics optimisation models
 +
* Finance optimisation models
 +
* Economic Models
 +
* Finance Models
 +
** http://en.wikipedia.org/wiki/Mathematical_finance
 +
** [[Linear Regression]]
 +
** [[Moving Average]]
  
 
==Examples==
 
==Examples==
 +
* Trading Models
 +
** http://www.thestreet.com/story/10414044/1/how-to-build-your-own-trading-model-in-8-steps.html
 
* http://www.forexnet.lv/EditModule.aspx?tabid=590&mid=1044&def=News%20Article%20View&ItemId=718
 
* http://www.forexnet.lv/EditModule.aspx?tabid=590&mid=1044&def=News%20Article%20View&ItemId=718
 +
* http://blogs.wsj.com/numbersguy/grading-the-forecasts-of-experts-182/
 +
* http://blog.centage.com/
 +
* http://en.wikipedia.org/wiki/Forecasting#Categories_of_forecasting_methods
 +
* http://en.wikipedia.org/wiki/Forecasting
 +
* http://www.ipredict.it/ForecastingMethods.aspx
 +
 +
 +
{{DISQUS}}
 +
 +
[[Category: Predictive Analytics]]

Latest revision as of 22:44, 29 August 2016

Forecasting is the process of making statements about events whose actual outcomes (typically) have not yet been observed. A commonplace example might be estimation of some variable of interest at some specified future date.

Predictions of outcomes based on data and prediction models. Forecasting involves Predictive Modelling.

There are millions of datasets out there. There are thousands of models. They determine the quality of forecasts. Reducing error is always an issue.


There are various forecasting methods listed below:

Extrapolative Methods

Extrapolative methods are to infer or estimate by extending or projecting known information.

# Method Description Actuary Usage Business Usage
1 Moving Average, Simple Moving Average Frequently used Widely used
2 Exponential Smoothing
  • Holt-Winters methods
Generally not used Widely used for TS analysis
3 Auto Regression Moving Average (ARMA) Generally not used Widely used

Explanatory Variable Methods

Use of a variable that explains a related response variable. Often called an independent variable or predictor variable.

# Method Description Actuary Usage Business Usage
1 Regression Analysis
  • Technical Analysis - Finance
Widely used Widely used
2 Predictive Modeling Gaining popularity Widely used
3 Artificial Neural Networks Generally not used Sometimes used
4 Econometric Modeling Generally not used Widely used

Simulation Modeling Methods

# Method Description Actuary Usage Business Usage
1 Cell-based Modeling Frequently used Frequently used
2 Systems Dynamics Simulation Generally not used Becoming more widely used
3 Multi-agent Simulation Generally not used Becoming more widely used


Judgmental Methods

  • Composite forecasts
  • Surveys
  • Delphi Method
  • Scenario building
  • Technology forecasting
  • Forecast by analogy

Composite Methods

# Method Description Actuary Usage Business Usage
1 Bayesian Forecasting Generally not used Generally not used
2 Others? Generally not used Generally not used


Forecasting

People


Forecasting Models

Every industry has its "models". These range in the thousands.

Examples


blog comments powered by Disqus