Do we need to have all the features in the same format like binary format?

Question

Recently I started using Weka Explorer for classification problem. I have two types of features

1. around 1200 features has binary features with values {0,1}
2. around 30 features has some numeric integer value like 1213.

Currently I am using naive Bayes classifier.

My question is: for better classification result do I need to have all the features in the same format like binary format?

Right now I am using weka unsupervised filter NumericToBinary to convert all numeric features to binary ones. Is this a correct way?

Any guidance will be helpful.

Answer 1

Weka explorer interface has quick way of determining your classifier's capabilities. Choose your classifier and click on its name. From that interface click to capabilities button see below image.

Naive Bayes capabilities are below.

CAPABILITIES Class -- Nominal class, Binary class, Missing class values

Attributes -- Numeric attributes, Binary attributes, Missing values, Empty nominal attributes, Nominal attributes, Unary attributes

Additional min # of instances: 0

As you can see it can handle numeric attributes therefore you do not need to use filter to change numeric attributes to binary. use this filter if it increases your classifier's performance.