i was making a classification model ( 3 class ) for early detection of cracks in ball bearings, the data set is limited 120 rows and 14 features. the classifiers and their parameters is listed below can you please suggest which model will be the best (not simply accuracy also consider model complexity )
I think it is better to try several models and see which one fits your data the most. You can't rely 100% on any suggesttions because small changes in the data may cause different results.
This is a difficult question. If you're interested in intuitive interpretation of your model ergo the results, you have to stick with the decision tree, but if you want to get the best predictive model for your classification task, the complex model/s would be the better choice. The performance of your (decision tree) model is actually pretty good, despite the size of your data, so stick with the simple model.
What are the 3 classes and the 14 features? How do the classifiers link the classes to feature space? Is there a causal explanation for this link?
I would go about answering these questions for myself in order to answer your question.
Hi, Dan Harris
thank you for answering
the 3 classes are bearings with low fault, medium fault and high fault.
14 features include wavelet based features, statistical features from the time domain data
I go by what kadhem said. Depending on your aim and objectives, try several models and select the right algorithm that fits your data the most.
It is better to use 10 fold cross-validation mode for calculating results and comparing with other trees.
Most probably if this is vibration data, random forest tree always exhibits superior performance.
- Badges
- Science topic