How to solve

$5$ Implementing Naive Bayes $[20$ pts$]$

You will now learn how to use Naive Bayes Algorithm to solve a real$-$world problem: text categorization. Text categorization $($also referred to as text classification$)$ is the task of assigning documents to one or more topics. For our homework, we will use a benchmark dataset that is frequently used in text categorization problems. This dataset, Reuters$-21578,$ consists of documents that appeared in Reuters newswire in $1987.$ Each document was then manually categorized into a topic among over $100$ topics. In this homework, we are only interested in earn and acquisition $($acq$)$ topics, so we will use a shortened version of the dataset $($documents assigned to topics other than "earn" or "acq" are not in the dataset provided for the homework$).$ As features, we will use the frequency $($counts$)$ of each word that occurred in the document. This model is known as the bag$-$of$-$words model and it is frequently used in text categorization. You can download Assignment $2$ data from the Canvas. In this folder, you will find:

train.csv: Training data. Each row represents a document, and each column separated by commas represents features $($word counts$).$ There are $4527$ documents and $5180$ words.

train labels.txt: labels for the training data

test.csv: Test data, $1806$ documents and $5180$ words

$6$

test labels.txt: labels for the test data

word indices: words corresponding to the feature indices.

Implement Naive Bayes Algorithm $($You cannot use any library function that implements Naive Bayes Algorithm$).$ Train your classifier on the training set that is given and report training accuracy, testing accuracy, and the amount of time spent training the classifier.