- ☀️Daily Log:
- Metadata
- Summary:
- By the end of 2021, 56 countries had registered more deaths per capita from COVID-19 than Sweden, a country that had not imposed lockdown mandate
- There seems to be a lack of retrospective on Sweden's policy and if that was the right choice
- There is an overestimation of the death toll caused by COVID-19, and compared to things like hunger it was much less severe
- Notes:
- This rosy figure is not so rosy when compared against other nordic countries that did go into lockdown
- Also it doesn't consider the demographic that it affects and it disproportionality kills people who are healthy
- Metadata
- Summary:
- One should increase the rate of experiment in order to arrive at the results
- Accepted fact that we don't know about the customer
- Provide basic tools for setting up an experiment
- Determine what sample size is needed for 80% power and 99% stat sig level for a particular kind of experiment
- Standard set of metrics
- And avoid pitfalls like sample ratio mismatch
- Standard dashboards, analysis and reporting structure
- It pays off to have standardized analysis
- User bucketing
- Having equal user buckets which have similar baseline performance
- Clean control group
- data hygiene - making sure user groups do not have any residual effect, doing A/A experiment
- Track the rate of experimentation and set groals around it
- Notes:
- Retrospective::
- One week ago: [[April 12th, 2022]]
- One month ago: [[March 19th, 2022]]
- One quarter ago: [[January 19th, 2022]]
- One year ago: [[April 19th, 2021]]