Naïve Bayes classifier

The naïve Bayes classifier is a simplistic implementation based on the Bayes formula. It is based on simple empirical and conditional probabilities, as evidenced in the actual data. Beyond the simplest assumption of observation independence, we don't have any restrictions in using this model.

Naïve Bayes for hypothyroid classification

A naïve Bayes classifier is fit using the naiveBayes function from the e1071 R package. The prediction and accuracy assessment is carried out using two functions, predict and sum:

> NB_fit <- naiveBayes(HT2_Formula,data=HT2_Train)
> NB_predict <- predict(NB_fit,newdata=HT2_TestX)
Warning message:
In data.matrix(newdata) : NAs introduced by coercion
> NB_Accuracy <- sum(NB_predict==HT2_TestY)/nte
> NB_Accuracy
[1] 0.9732704403

The accuracy of the naïve Bayes classifier is 97.33%, which is the same as the logistic regression model and less than the one provided by the neural network. We remark here that it is only a coincidence that the accuracy of this method and logistic regression is the same.

上一章目录下一章