I want to experiment with Machine Learning algorithms and compete in Kaggle.com. Can you please suggest any probabilistic classifiers in Machine Learning, their advantages and disadvantages?
If you want probabilistic classifiers, Naive Bayes classier or GMM can be possible choices.
Dear Kaleab Tekle,
Look the link, maybe helpful.
https://pdfs.semanticscholar.org/6c8d/009fe19eda2966a8509865225892bdcde90c.pdf
http://www.isaet.org/images/extraimages/P1215028.pdf
Article Combining Probabilistic Classifiers for Text Classification
Regards, Shafagat
Since you need a probabilistic clssifier you may experiment with KNN; if not SVM
"Machine learning is not a set of algorithms. The algorithms are limit, but the real world tasks vary. There must be some tasks you cannot solve if you just use these existing algorithms without change. Therefore, we should not only focus on the derivation of algorithms and their programming implementations. The ideas behind algorithms are the keys to modify the algorithms to fit the reality tasks.
There is no universal best machine learning classifier. Every machine learning approach has an inductive bias. Therefore, for any classifier, there exists some data distribution where it performs worse than another classifier."
https://www.quora.com/Is-there-a-best-machine-learning-classifier
In machine learning, a probabilistic classifier is a classifier that is able to predict, given an observation of an input, a probability distribution over a set of classes, rather than only outputting the most likely class that the observation should belong to. Probabilistic classifiers provide classification that can be useful in its own right or when combining classifiers into ensembles.
From the theoretical point of view KNN (K-Nearest Neighbors) is a perfect classifier. You can prove that. The problem is that as the dimensionality of the data increases, KNN just starts being less effective.
There isn't a single "best" classifier per se, it usually comes up to your data, and many times, the way you preprocess you data is more important than the classifier you use. At the end of the day, the classifier is going to be as good as the data you put in it.
No, there is no all inclusive best machine learning classifier. Each machine learning approach has an inductive predisposition. Subsequently, for any classifier, there exists a few information circulation where it performs more regrettable than another classifier.
regards
Agree with @Hassan Nima.
Follow this link maybe helpful
https://pdfs.semanticscholar.org/6c8d/009fe19eda2966a8509865225892bdcde90c.pdf
regards
What if we want to predict numerical values such as probability of sales of items, do we use probability classifiers or predictors?
- Badges
- Science topic