Naive Bayes?, KNN? and C-Means?
i.e. I'd like to compare these classifiers, so can anyone help?
I attach a very similar question that was posted yesterday. I refer you to the answers there (including mine), where you can find some papers that compare classifiers on the base of several criteria, including accuracy.
Additionally to that, I would also recommend some papers on statistical comparison of classifiers, e.g. "Statistical comparisons of classifiers over multiple data sets" or "Approximate statistical tests for comparing supervised classification learning algorithms". These are rather useful for analyzing the significance of differences in accuracy (however calculated) between classifiers.
https://www.researchgate.net/post/Is_there_a_comprehensive_survey_on_what_machine_learning_technique_performs_best_on_various_kinds_of_problems2
I attach a very similar question that was posted yesterday. I refer you to the answers there (including mine), where you can find some papers that compare classifiers on the base of several criteria, including accuracy.
Additionally to that, I would also recommend some papers on statistical comparison of classifiers, e.g. "Statistical comparisons of classifiers over multiple data sets" or "Approximate statistical tests for comparing supervised classification learning algorithms". These are rather useful for analyzing the significance of differences in accuracy (however calculated) between classifiers.
https://www.researchgate.net/post/Is_there_a_comprehensive_survey_on_what_machine_learning_technique_performs_best_on_various_kinds_of_problems2
Every Machine Learning technic have its pros and cons. You can try with some subset of your training set and do folding for testing with several classifiers. Depending of the functions that the classificacion represent, are going to be best one classifiers over the otherones.
In general, there is no single technique that can outperform all other techniques in all cases. but you can select the best classifier based on the characteristics of the data you have. This is a useful paper:
Supervised Machine Learning: A Review of Classification
Techniques
https://s3-us-west-2.amazonaws.com/mlsurveys/54.pdf
A very useful paper:
https://s3-us-west-2.amazonaws.com/mlsurveys/54.pdf
BTW... No Free Lunch Theorem for Optimization (to build a Classificator it is like a optimization problem). There is no a "Best" method to build one for all cases. http://ti.arc.nasa.gov/m/profile/dhw/papers/78.pdf
A important point that is to be considered are the types of attributes of data, that can influence of algorithm performance. the paper http://www.cs.waikato.ac.nz/~mhall/HallHolmesTKDE.pdf presents a benchmark comparison of several attribute selection methods for supervised classi cation
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining