I know my question is so generally it depends on data set but my question is performance of an data mining algorithm depends on structure of data set for example size of dataset،number of feature،missing values،outliers،type of feature ،synthetic data and ،.... I want to know these structural metrics if know some book or paper about this subject please inform me .
I know about performance metrics like accuracy , recall and .... but my question is not aboute these metric I want to know for example why c4.5 performs well in this data set or what element in this data set cause this result
Hello,
I think, your question is "Which Factors Affecting Classification Algorithms Performance". Please check following sources:
Url 1: https://pdfs.semanticscholar.org/0b79/fa21c9fd3df333852eb4c92530e4faf267b7.pdf
Url 2: http://eacharya.inflibnet.ac.in/data-server/eacharya-documents/53e0c6cbe413016f23443704_INFIEP_33/4/ET/33-4-ET-V1-S1__factors_classification.pdf
Maybe, it helps
Thanks
Hi Saba,
Your question is very interesting and also very common. For a long time it made me think. Unfortunately, I think that there are no general rules and therefore a single response to it. Learning algorithms seem to have different souls depending on the world where they live. For this reason it is important to keep in mind that the knowledge discovery process is an iterative and incremental process, where the phase of data understanding and preparation is very important. I advise you to learn more about the CRISP-DM methodology, if you don't already know it.
Pete Chapman (1999); The CRISP-DM User Guide.
Even the papers suggested by Ahmed Iqbal can help you.
Good job,
Gaetano
Thank Gaetano Zazzaro and Ahmed Iqbal for answering.your answers will help me in doing my bachelor project
- Badges
- Science topic