Intro to Data Analysis, General Assembly at SXSW 3/10/18
Presenter: Bianca Mounce, Forward Together
What is Data Analysis? Transforming raw data with the purpose of drawing conclusions. These conclusions don't need to be solidly "conclusive", but should be digestible. "My recommendation is that I need more data"...I want to append data to that dataset to understand more.
Steps for transformation process of data analytics workflow
1. Identify the problem: need to be strategic in asking the right question
Who are our customers? What do they care about?
Parse the question:
Scope (geographical)
What (demographic of who is impacted)
2. Obtain the data: communicate clearly with partners what data we need
3. Understand the data: can correctly interpret the results and that they're credible
Budget in time in your workflow for this step when you don't know the format it'll be received in
4. Prepare the data: standardization and cleaning up dupes, etc. consistency
Consequences to this on both large and small scale
Pay attention to outliers in your data...could there be an issue with the data?
"Narrow"
"Scores": may be created/determined by Data Scientists
5. Analyze the data: organic and dependent on the question
Can start w/ highs and lows to pique curiosity
"Who should we call?"
6. Present the results: Be the translator of the numbers
Know your stakeholders, what information they care about
Tells the story of our data
Show success rates, attribute the successes to the data analysis
TOOLS
- Mastery of Excel
- SQL
- Tableau
- Data.World
- Metabase
Tip: On a spreadsheet, helpful to write questions we want to answer
Questions asked: How do you determine if your dataset is large enough?
When working with external organizations, what are the challenges of agreeing upon the question 1?
I'm sure in each business model it is different, but who determines the most effective way to present the data for either your external or internal stakeholders?
How much domain knowledge do you need to be a successful analyst?