Data Analyst

For data analysis, I think the most important thing is to clarify the business goals, identify the gaps with the goals, and determine the problems that need to be solved.

Core Issue 1: Defining the problem - Identifying the reasons for the gap (requires analytical ability and business sense, both personally and on average for peers)

Core Issue 2: Cause analysis - After finding possible reasons, refine the indicators

Core Issue 3: Analyzing indicators - Find the necessary data to support the analysis (how to find it depends on how to utilize tools)

Core Issue 4: Data analysis - Analysis methods

Core Issue 5: Solution testing - Once the cause is hypothesized, how to judge if the assumption is correct and proceed with the experiment

Core Issue 6: Continuous improvement - How to find predictive indicators to prevent similar problems from recurring and detect them early

For example, when analyzing the overspending of advertising budget:

First, you need to realize that the problem you are analyzing is overspending. Did you have a definition for overspending previously?

Then, what are the reasons?

  • Sales
  • Price
  • Conversion
  • Bidding
  • Increase in bidding

Overall increase? Local increase? factors? Market factors?

A particular department is having a problem; which product in the department?

How to prevent it?

Why do it?

What to do?

How to do it?

As for SQL statements, such as:

  1. Execution order
  2. Table additions, deletions, and reductions, primary key PID
  3. Aggregate functions
  4. Date conversion
  5. Table joins
  6. Window functions
  7. Sliding window functions
  8. Row and column transformations

In the era of GPT, is this really important?