katcorr
commented
4 years ago What an exciting plan, team! Ambitious with lots of questions and data sources -- but good to see you've already identified specific data sources with the data needed. Appropriate incorporation of spatial data and text analysis in a creative way.
- from the rtweet github page: "{{rtweet}} should be used in strict accordance with Twitter’s developer terms." please be sure to review those terms to confirm you're abiding by them!
- "Our project will be in the form of a shiny app," -- clarification that you do need to create the blog post (with writing that provides context and analysis, limitations etc.) but you can certainly include a link to your shiny application in the blog post and have that shiny app incorporate the meat of your project (all the visualizations etc.)
I'm really looking forward to your blog post!
Update 1: 10/10
pbmorais
mattadams23
Group Name: Let’s Get Fiscal Group Members: Pedro Morais, Braedon Fiume, Matt Adams October 30, 2020
Update #1
General Idea: The COVID pandemic has forced many businesses to shut down, however, probably the most successful restaurants are those of fast-food models. Their significantly less required human contact and quick delivery are complemented by the restrictions of COVID. Following this, it would be interesting to examine the patterns of the most successful fast food chains using spatial data and text analysis. Using spatial data, we can show population densities and the amount of fast-food chains already located in areas. Using text analysis, we can show the most recent consumer sentiments of the restaurants in those specific locations. Then, we can narrow down a list of plausible and beneficial locations for the new stores to open, creating sustainable employment opportunities and stable cash flows for cities. Some of the questions we hope to address are: What are the 5 fast-food restaurants that have the most establishments in the United States? What geographical locations have the most or least amount of restaurants per capita? In specific locations, what proportion of consumers feel positive or negative about the fast food restaurants (via Twitter)? What cities most need income and employment opportunities?
Data:
This project requires significant information, and thus there are a variety of data sources that are needed. In our past project, our team ran into many complications with data scraping and infinite loading. However, this project has datasets that are easily accessible through Kaggle, a trusted data science company, and R packages. To tackle the spatial data visualizations of our project, there are three critical Kaggle datasets that can be downloaded manually through a csv file. The first is the dataset containing a list of 10,000 fast-food restaurants and their location with longitude and latitude locations. The second dataset from Kaggle contains a list of 754 cities across the United States, and their corresponding population as of 2016. Finally, there is a dataset that contains the U.S household income statistics of cities across the United States, allowing us to view cities that are in need of income. For text analysis, we will use a library from R called “rtweet”. This library and its packages are amazing, allowing us to request 18,000 tweets from the past 9 days based on certain words, like “#McDonalds” or “#Wendys”. Combining “rtweet” and the “get_sentiments” function, we can use text analysis and proportions to see the amount of positive and negative tweets for the top 5 fast food restaurants, as well as positive and negative tweets for specific locations! All these datasets will provide insight into recommending a geographical location for a chain's next restaurant location.
Resources: Fast-Food Restaurant Dataset: https://www.kaggle.com/datafiniti/fast-food-restaurants?select=FastFoodRestaurants.csv City Population/Densities: https://www.kaggle.com/mmcgurr/us-city-population-densities City Household Income Statistics: https://www.kaggle.com/goldenoakresearch/us-household-income-stats-geo-locations?select=kaggle_income.csv R Package “rtweet” Explanation: https://www.earthdatascience.org/courses/earth-analytics/get-data-using-apis/use-twitter-api-r/
Final Product: This project will be incorporating spatial data and text analysis to provide consultation for the top 5 fast food restaurants in the United States. We will be displaying spatial data visualizations of the amount of restaurants the fast food chains currently have across the country, as well as the population densities and wealth of the cities across the country. In addition, hopefully we can implement the text analysis of twitter into spatial data, representing which locations show the most positive and negative sentiments for the fast food chains. Examining all this information and data visualizations, hopefully we can recommend a suitable location for each of the top 5 fast food chains next restaurant. Our project will be in the form of a shiny app, and the user will be able to interact with the plots to examine the relationship between a specific restaurant and their locations and consumer sentiments across the country. The general population density and wealth of cities will be found on one singular tab for reference.
Schedule:
Week 1 (November 1): Tuesday, November 3: Have datasets downloaded and consolidated, have initial sentiment analysis data gathered. Thursday, November 5: Finalize tweet scraping and sentiment analysis data, begin initial data spatial visualizations for fast food chain distribution.
Week 2 (November 8): Tuesday, November 10: Create visualizations for sentiment analysis. Thursday, November 12: Create app for recommending the next suitable location for top 5 restaurant food chains. Saturday, November 14: Begin uploading graphics and formatting website.
Week 3 (November 15): Monday, November 16: Finalize write-up for website and prepare for presentation. Tuesday, November 17: Present and blow everyone’s mind!