Is Naive Bayes a good classifier?

Question

Here is an example of training a model using the Naïve Bayes classifier on the Glass dataset(from UCI). The objective is to predict the type of glass based on the 9 parameters. The metric used to understand the classification result are confusion matrix and classification report.

The program is available here

Few Observations/ Questions

1. By Varying the ‘random_state’ value inside the function train_test_split, we can observe different accuracy values? Is the behavior correct?
2. The StratfiedShuffle method of  train_test_split also produces random results on the every run. Is there a bug with Naïve Bayes classifier implementation?