Homework 1: Data Analysis

Step 1: Parse the Data

Files: analysis.js

Your first task will be to utilize two dependencies, the built-in fs, and papaparse, which you installed earlier, in order to turn the CSV data into an object we can work with in JavaScript. Here are the specifications and tips for using each of these dependencies:

At the end of this section, you should have your data stored an object in a variable.

Step 2: Data Cleaning

Files: analysis.js

Now that you have your csv data parsed, you will want to destructure and clean your data in preparation for analysis. Data you recieve from large datasets may not always be clean- in fact, this dataset has many null values, including null review texts, ratings, countries, genders, and app versions.

As mentioned earlier, papaparse converted all the data into strings. You will be converting all of the column values into their proper types. Additionally, you will restructure all the user properties into its own object property for each review.

Your data cleaning goals are as follows:

• Filter out every record with null column values, except user_gender, as a null gender value is allowed.
• Merge all the user statistics, including user_id, user_age, user_country, and user_gender, into an object that holds them called user, while removing the original properties.
• Convert the review_id, user_id, num_helpful_votes, and user_age to Integer
• Convert the rating into a Float
• Convert review_date into a Date
• Convert verified_purchase to Boolean

Here is an example of one of the cleaned, filtered records from the dataset:

{
  review_id: 99,
  app_name: 'Grammarly',
  app_category: 'Music & Audio',
  review_text: 'Stupido malattia donna 
    magari già posare sbagliare qualità. 
    Tempo vino morale sviluppo 
    ora popolazione avvicinare.',
  review_language: 'fi',
  rating: 2.3,
  review_date: 2024-06-28T22:59:58.000Z,
  verified_purchase: false,
  device_type: 'Windows Phone',
  num_helpful_votes: 1090,
  app_version: '6.9.40-beta'
  user: {
    user_age: 44,
    user_country: 'Vietnam',
    user_gender: '',
    user_id: 9262579,
  }
}

At the end of this section, you should have an object that contained your filtered and cleaned data. Make sure your properties exactly match the property names from the original dataset and have the correct data types stored in them.

Step 3: Sentiment Analysis

Files: analysis.js

With newly cleaned data, we are now ready to start analysing it! In this step, we will add a property to each record called sentiment that represents a general label of the rating each user granted to apps. We will also destructure some parts of the reviews for analysis later on.

Write a function called labelSentiment() that takes in a rating as an argument and returns strings based on that rating:

• positive if the rating is greater than 4.0.
• neutral if the rating is between 2.0 and 3.0.
• negative if the rating is below 2.0

Next, we will sort the sentiments by app, and then by language, into arrays of objects. Do these analyses in the functions sentimentAnalysisApp() and sentimentAnalysisLang() respectively. Be sure to follow the proper object format for returning arrays of these statistics provided in the function documentation.

For this section, you may find destructuring your data into other objects to be helpful in sorting your data for analysis and printing. Refer to the lecture slides for how you can destructuring and other syntactical techniques for this section!

Step 4: Summary Statistics

Files: analysis.js

To wrap up our analysis, let us look at some basic summary statistics. Using the cleaned data and any objects you have created from the previous section, answer the following statistical analysis questions:

• What is the most reviewed app in this dataset, and how many reviews does it have?
• For the most reviewed app, what is the most commonly used device?
• For the most reviewed app, what the average star rating (out of 5.0)?

Create an object to store the answers to these questions and return them from the function summaryStatistics(). Use the object format provided in the homework documentation.