Comparing to Historical Averages
Last modified: July 15, 2019
Contribute to Data School
We are actively working on this chapter. We are looking for two types of contributions:
- Help writing this chapter
- Share your story about working through this problem at work
Please reach out to @Matt David on our slack to discuss what you want to contribute.
Comparing to Historical Averages In Progress Overview
We work with hundreds of companies who are trying to get valuable insights out of their data. We see companies comparing the success of their product or feature as compared to a historical average.
We want to show how comparing to a historical average can be misleading. Is the past 6 months an appropriate window? Were there any trends in the data prior to our test? Then we want to cover how to create SQL queries to compare to historical statistics. Lastly we want to walk through interpreting historical statistic comparisons.
Outline + Learning Objectives
The problem with historical averages
- Describe why a historical average can be misleading
- Describe how it may hide a trend in the data
Create appropriate historical comparisons
- Create SQL queries to compare against previous set of months
- Create SQL queries to compare against the same month last year
Interpret historical statistics
- Differentiate between analyses that use historical comparisons appropriately and ones that do not
Written by:
Reviewed by:
