Problem Description:

There are at least two ways to store/maintain training dataset for the purpose of using Naive Bayes Classifier:

• Store the raw training data items. The advantage is that add a new training data item is extreamly easy. The disadvantage is that the conditional probability will needed to be calculated from counting (traverse every data item) when a query needs to be classified (time inefficient) and it needs a lot of storage space to store all the raw data (space inefficient).
• Store the conditinal probability table. The advanteage is that query classification can be done very efficiently. The disadvantage is that adding a new training data item becomes very difficult.

Your tasks:

Using the data set CarData.csv used in assignment 3 as an example, design a data structure and related algorithms to make it possible that a program can both maintain incrementally added training data items and perform the classification task using Naive Bayes Classifier in time and space efficient ways.