Hi,
The general data mining process is composed of different steps from data collection to KDD presented in incremental way on data mining lectures. I need to know what means exactly the returning arrows going from KDD to data collection (as you can see in the image: http://www2.cs.uregina.ca/~dbd/cs831/notes/kdd/kdd.gif).
I need to know more details about it, and if it's possible please show me a concrete example.
Thank you.
the returning arrows helps in refinning the selected data (selecting methods or parameters) from the initiel big mass in order to get more accurat information (i.e. when accuaring new knowledge we have new information that could affect the process of selecting data in the first place )
The final layer generates the result/label. If it is correct then no cost or in other case result is penalized in terms of increase or decrease weights. So that next time the result reaches more close to actual. Ultimately learning in NN is all about weight adjustments.
After you obtain the results from KDD, you interpret and evaluate them and you may find that the results are not so good or that they do not answer what you was looking for. Therefore, you realize that you have to change something in the previous steps and start the analyses from the beginning: perhaps to select different data, perhaps preprocess or transform the data in some other way etc. So the arrows do not mean that you somehow update the initial data - they simply mean that very often you should return to the beginning because you are not satisfied with the results.
- Badges
- Science topic