search

penguin-statistics / backend-next

The refactored Penguin Statistics v3 Backend. Built with Go, fiber, bun and go.uber.org/fx. Uses NATS as MQ and Redis as state synchronization.
MIT License
176 stars 19 forks source link

"Report Count as X-axis" mode in Trend #47

Open GalvinGao opened 2 years ago

GalvinGao commented 2 years ago

For an activity that is just opened, the users care more about how will the drop rates of this item in the current activity be. To give them the fastest way to understand this and also give a better way of how the samples finally rest around a particular rate.

Besides only reporting the count of accumulated quantity, an accumulated standard_deviation should also be calculated so that visualization on confidence intervals could be easily visualized, assuming the drop rate across different stage clears is a binomial distribution.

FlandiaYingman commented 2 years ago

Also showing the confidence level to users might be a good choice.

Some of the users mistakenly assumed that our site is an authority. And they mistakenly trusted the statistic even if the sample size is extremely small. Showing the confidence level to them could hint that the drop rate is a statistical number and could help users to decide whether trust the statistics or not.

This also prevents our site from being blamed because of the inaccuracy of drop rates, like, by users from bilibili or elsewhere. LOL