Planning and Designing Your Project
ISTA 498
Goals of the worksheet
A bit of thought now can save a lot of pain later
Provide a clear roadmap to completing your project
Make sure skills/ideas are in order
Make sure tasks are delegated equally throughout the group
Tentative project title
Should be catchy but informative
If someone sees only the title on your resume, will they click it?
Future.Gaming.exe
Catchy, but not informative…
State Farm Social media Sentiment Analysis
Informative, but not catchy…
What are you solving/answering/achieving?
Essentially, why should I care about your project?
When you lead talking about your project, you start with this problem!
State Farm Group
You’re goal isn’t a sentiment analysis – that’s just a technique
Goal is to understand public perception of a new product by leveraging social media data
What is the minimum endpoint
What are the minimum items you need to call your project a success?
State Farm group
Corpus of tweets relating to State Farm in general as well as the specific product
Sentiment analysis to compare attitude
Quality visualizations that clearly display sentiment dimensions
Presentation that you’re willing to present to State Farm employees
This tells you where you need to get to and helps guide immediate needs
Can add stretch goals later
What are the computational skills and resources
Need to ask yourself what you’ll need to achieve your endpoint
State Farm Group
Dev Twitter account
Understanding of APIs and how to gather tweets via twitter’s
Data wrangling in python or R
Sentiment Analysis
Data Viz – ggplot
Great presentation skills
What data are needed
Given your project, what data are needed to reach endpoints?
State Farm
Tweets that focus on two areas – State Farm in general and specific product
Ideally tweets starting from when product was released
Clearly exists on twitter
But how much can you get?
Is the format easy to use or will it need extensive wrangling?
What are the pain points? New Skills?
Identifying potential problems and points of learning is critical to making a timeline later
State Farm group
How much twitter data can you get through free API?
Can you divide queries among people or over time?
What to do if not?
Do you know how to access via API or packages?
What are the specific tasks done by each member?
You don’t want to have 3 of 5 people doing one simple task
State Farm Project
Member 1 – Twitter API access and getting tweets
Member 2 – Sentiment analysis
Member 3 – Viz
Can and should have overlap!
Don’t silo yourself in one area
Member 1 should know what 2 is doing so they can make sure they’re getting the right data
Member 3 should be working with 1 and 2 to ensure own data stream and accurate representation of data
Timeline and how to maximize efficiency
Don’t need to structure it a linear way
What can you do first that will allow whole project to get started
State Farm
Get some tweets in some method
Can start coding sentiment analysis
Can start data wrangling and viz
You should set your endpoint at the presentation showcase and work backwards
Last minute work is not acceptable
Feedback from site supervisor
Once the form is completed you need to have your site supervisor approve it. They should be directed to give feedback on the bottom.
This sort of managerial feedback is critical as they’ll be able to quickly identify the feasibility of a project given they’ve been doing it for a while
You want this to be detailed enough that they can provide constructive feedback.
To vague and they will only be able to hint general issues which might not be helpful.