Project: Data Wrangling with Json SOlution

1 Background




JavaScript Object Notation (JSON) is a format for storing and exchanging data. It originated from an earlier version of JavaScript and is now supported by almost every programming language. JSON is designed to be human- and machine-readable. For example, the following text is JSON:







[




{




"Department": "Computer Science",




"Course Number": 220,




"Instructor": "Joydeep Biswas",




"Offered": true




},




{




"Department": "Computer Science",




"Course Number": 589,




"Instructor": "Joydeep Biswas",




"Offered": false




}




]




This example shows a JSON array with two elements, where each element is a JSON dictionary. Each JSON dictionary has a set of keys—which are strings—and values—which are JSON values themselves. The example shows that JSON values include strings, numbers, and booleans. Note that arrays and objects are JSON values themselves. Therefore, arrays and dictionaries may be nested:




[




{




"x": {




"y": [ 1, 2, 3 ]




}




},




"Something else"




]




The example above also shows that JSON arrays may be heterogeneous :the two elements of the array are an object and a string.




Ocelot has a function called lib220.loadJSONFromURL that takes a URL for a JSON file as a string and returns the file as a JavaScript object. This function also takes care of parsing the




JSON string notation, and providing you directly with an object. Here is an example of how to call lib220.loadJSONFromURL() , as executed in the Ocelot terminal:




lib220.loadJSONFromURL('https://people.cs.umass.edu/~joydeepb/yelp-tiny.json'); < [



{




name: "China Garden",




city: "Stanley",




state: "NC",




stars: 3,




review_count: 3,




attributes: { RestaurantsAttire: "casual", Alcohol: "none", OutdoorSeating: false




}, categories: [

"Chinese",




"Restaurants",




]




},




{




name: "Enterprise Rent-A-Car",




city: "Mesa",




state: "AZ",




stars: 4,




review_count: 3,




attributes: {},




categories: [ "Hotels & Travel", "Car Rental",

]




},




]



















2 The Yelp Dataset




The business review site Yelp releases a large dataset of restaurants (and other businesses) in a JSON format. In this assignment, you will use this dataset to answer vital questions such as “What is the most popular restaurant in California?”. Unfortunately, the full dataset is nearly 7GB, which is too large. Therefore, you will instead use a subset of the Yelp data. Each entry in

the dataset is a JSON object that looks like this:







{




name: "China Garden",




city: "Stanley",




state: "NC",




stars: 3,




review_count: 3,




attributes: {




RestaurantsAttire: "casual",




Alcohol: "none",




OutdoorSeating: false




},




categories: [




"Chinese",




"Restaurants"




]




}




The following url can be used to load the JSON file for this dataset into Ocelot:




https://people.cs.umass.edu/~joydeepb/yelp.json







3 Programming Task







Overview




The goal of the programming task is to define a class FluentRestaurants that supports the fluent design pattern to filter the dataset. We can use this class to perform queries such as “What vegan restaurants are in Wyoming?” Or “Which Mexican restaurants in NY are rated below 2 stars?” The fluent design thus allows the queries to be chained in arbitrary orders, with specified constraints, much like a user might wish to, on the Yelp website, to find specific restaurants of interest. For example, here is how you would use the class FluentRestaurants to run two queries:







let data = let f = new

FluentRestaurants(data);lib220.loadJSONFromURL(' https://people.cs.umass.edu/~joyd




eepb/yelp.json' );




f.ratingLeq(5)




.ratingGeq(3)




.category('Restaurants')




.hasAmbience('casual')




.fromState('NV')




.bestPlace().name;




f.ratingLeq(4)




.ratingGeq(2)




.category('Restaurants')




.hasAmbience('romantic')




.fromState('AZ')




.bestPlace().name;










The key idea is that you can compose these functions together to pose complex data queries. In the above snippet, the first query determines the best “casual” restaurant in Nevada with at least 3 stars and at most 5 stars. The second query determines the best “romantic” restaurant in Arizona that has a rating of at least 2 stars and at most 4 stars.










Specifications




Create a JavaScript file in Ocelot, define a FluentRestaurants class, and then implement the class methods enumerated below utilizing the Fluent Pattern discussed in lecture. The constructor should be defined as follows:







constructor(jsonData) {




this.data = jsonData;




}







To assist with JSON objects, you will need to use the lib220.getProperty(jsonObj, memberStr) library function. lib220.getProperty takes in a parsed JSON object and the string name of an object member and returns another object. The returned object has two member variables, found and value ,to indicate whether the property with the specified string was found in the object or not, and if so, to return its value. Here is an example usage:







test("Usage for getProperty", function() { let obj = { x: 42, y: "hello"}; assert(lib220.getProperty(obj, 'x').found === true); assert(lib220.getProperty(obj, 'x').value === 42); assert(lib220.getProperty(obj, 'y').value === "hello"); assert(lib220.getProperty(obj, 'z').found === false);




});




The FluentRestaurants class must implement the following methods:

fromState(stateStr) ,that takes a string, stateStr ,and returns a new FluentRestaurants object in which all restaurants are located in the given state, stateStr .



ratingLeq(rating), that takes a number, rating ,and returns a new FluentRestaurants object that holds restaurants with ratings less than or equal to rating.



ratingGeq(rating), that returns a new FluentRestaurants object that holds restaurants with ratings which are greater than or equal to rating.



category(categoryStr), that takes a string, categoryStr, and produces only those restaurants that have the provided category, categoryStr .



hasAmbience(ambienceStr) ,that produces those restaurants that have the given ambience. Each restaurant record contains an ‘attributes’ key that may or may not contain an Ambience key, which itself is an object:



{ ...




attributes: {




... Ambience: { hipster: false, trendy: false, upscale: false, casual: false




}




}




}







Each member of the Ambience object has a key-value pair for ambience types, and whether the restaurant has that ambience or not.




bestPlace(), that returns the “best” restaurant. The “best” restaurant is the highest rated restaurant. If there is a tie, pick the one with the most reviews.



4 Testing Your Code




You will have to test your code thoroughly to ensure that it robustly handles the challenges of working with real-world data, including missing / optional fields and variable structure. To help you get started, we have provided a few test cases here. It is up to you to define additional tests to check your solution for correctness.







const testData = [




{




name: "Applebee's",




state: "NC",




stars: 4,




review_count: 6,




},




{




name: "China Garden",




state: "NC",




stars: 4,




review_count: 10,




},




{




name: "Beach Ventures Roofing",




state: "AZ",




stars: 3,




review_count: 30,




},




{




name: "Alpaul Automobile Wash",




state: "NC",




stars: 3,




review_count: 30,




}




]




test('fromState filters correctly', function() {




let tObj = new FluentRestaurants(testData);




let list = tObj.fromState('NC').data;




assert(list.length === 3);




assert(list[0].name === "Applebee's");




assert(list[1].name === "China Garden");




assert(list[2].name === "Alpaul Automobile Wash");




});




test('bestPlace tie-breaking', function() { let tObj = new FluentRestaurants(testData); let place = tObj.fromState('NC').bestPlace(); assert(place.name === 'China Garden');




});