The data includes reviews of some restaurants in Charlotte. Below is the list of

The data includes reviews of some restaurants in Charlotte. Below is the list of variables in the data:
ratingStar rating
textReview text
nameRestaurant name
review_countTotal Number of Reviews of the Restaurant
is_open1: The restaurant is still open, 0: Restaurant is closed
Step 2:First, we need to perform text analysis to construct some variables from the review text. We subsequently are going to use these variables in order to answer the questions.
Perform sentiment analysis of the reviews using the “analyzeSentiment” package.Note that the sentiment analysis could take a few minutes to be completed, it might event take longer on Apporto.
Create a word cloud of all the reviews and provide a short summary of what you find in the word cloud.You may change the min_count in the code to some readable word cloud.
Perform topic modeling (either LDA or BTM) using a selected number of topics.The number of topics is set to 10 in the code. However, based on the coherence of the topics, you should increase or decrease it, but make sure you justify your selection of number of topics.
Note that the topic modeling could take a few minutes to be completed, it might event take longer on Apporto.
After performing topic modeling, make sure you store (1) top terms in each topic in order to be able to label the topics, (2) topic assignment for each review. The first one should have been displayed as a plot in the R-Studio output, and the second one should be included in the output csv file.
Second, based on the variables generated, answer the followings questions by using either visual (plots) or numerical (tables) representation of the data in Radiant or any other software.
Let’s start by exploring to see if we can find any evidence for the argument that restaurants with higher ratings are going to have a higher number of reviews. (You may use the two variables of rating and review_count)
Next, let’s see if the sentiment of the reviews is related to the rating that people provide. This would be a test of our sentiment analysis because we expect people who rate lower also use negative words in their reviews, so sentiment should be lower too.
Finally, let see how the topics of the reviews are related to the ratings, is there a topic that can predict the ratings?
You may use regression analysis and choose the rating as the outcome variable and the topics as the predictors.
You may change the type of the topic variable into a factor under the Data -> Transform -> Change type -> as Factor and choose the topic variable.
Make sure you interpret the results using the sign and significance of the parameters.
Using the estimated coefficient, what are the top three topics that can predict the ratings?
Third, let’s evaluate the relationship between the following variables and the restaurant being currently open vs. close. You may use either a table or a plot to evaluate these relationships.
Sentiment score
Review count
Topics
Ratings
Consider the above explored relationships and discuss which one seems to be a better predictor for a restaurant being still open.
Step 3Submit the assignment to Canvas.
Step 4Conduct a Peer Review in Module 7 (see instructions below).