Figure 46 illustrates Modelling Processes in Context:
| Reference (or cite) Article | ||
![Kahlon, R. S. (2013) Modelling Processes [Online]. dkode: United Kingdom, England, London. [Published on: 2013-02-13]. [Article ID: RSK666-0000108]. [Available on: dkode | Ravi - https://ravi.dkode.co/2013/02/modelling-processes.html].](https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEi1QAw0HHVAsw96tyjOTY02IagXr-zhdVtJRyUDsAaICwZroxbpnwCX-zNgES2I34hXdkOxpikEtky4waS6ytDbEa4X1Xw6ty4HJkRFZbI9JJBnmrIw8HhyphenhyphenpQfz3nPPaSOyohjqQSeB2TeROMz4U20KQHWfj46x1sfWQ2JhIMTclZlaecIYCHqYYinU6g5s/s1600/reference-dkode.png "Kahlon, R. S. (2013) Modelling Processes [Online]. dkode: United Kingdom, England, London. [Published on: 2013-02-13]. [Article ID: RSK666-0000108]. [Available on: dkode | Ravi - https://ravi.dkode.co/2013/02/modelling-processes.html].") | Kahlon, R. S. (2013) Modelling Processes [Online]. dkode: United Kingdom, England, London. [Published on: 2013-02-13]. [Article ID: RSK666-0000108]. [Available on: dkode | Ravi - https://ravi.dkode.co/2013/02/modelling-processes.html]. | |
This modelling processes diagram is a very simple and explicit form. I'd like to introduce you to "Heteroskedasticity" (from Greek), in connection with your diagram but expressed and combined in statistical terms. Firstly, the definition of heteroskedasticity is meant by a sequence of random variables which contain different variances- see here. If we split this word down, hetero (different) and skedasticity (dispersion). There are two types of heteroskedasticity, (a) pure and (b) impure.
ReplyDeletePure arises if a model is correctly specified, however, the errors are heteroskedastic. Whereas where variables are omitted, here the model is most likely to be mis-specified, this is known as impure. Subsequently, the specification error induces heteroskedasticity.
At this point, on your diagram, the cycle of redefine is the refinement of the model and heteroscedasticity would be beneficial of softening the elements to make the model more efficiently and effective.