HowSuneIsNow
Regular
I'm interested in creating a statistical analysis package for excel where you could take your times from CCT or a similar program and be able to see your progress in more statistically rigorous and meaningful ways than just seeing what your average of 100 is. I would like to include a running average of times weighted by date with a confidence interval.
This would give you a graph of your sub-x score over time and take into account variance and sample size so you could see when you're actually making improvements. You would have measurements like the upper bound of a confidence interval of your true mean which in my opinion is way more useful than a sub-x score because it takes into account sample size and variance. To over simplify this. you could find what the upper bound of your true mean average of infinity solves would be even if you haven't made infinity solves. It's like a scientifically rigorous sub-x score. I could also have a graph of your best ao5,ao12, ao100, ao1000 over time. Another thing I thought would be interesting is to find a confidence interval for randomly sampling 5 solves and finding a tournament average of 5 and showing you the confidence interval for what score you would get at a tournament based on a solving session.
Does something like this already exist that I could add on to?
I have very little data of my own solves. I would love to have a excel file of someones solves to use when developing this.
I'm curious what kinds of features would be useful to people.
This would give you a graph of your sub-x score over time and take into account variance and sample size so you could see when you're actually making improvements. You would have measurements like the upper bound of a confidence interval of your true mean which in my opinion is way more useful than a sub-x score because it takes into account sample size and variance. To over simplify this. you could find what the upper bound of your true mean average of infinity solves would be even if you haven't made infinity solves. It's like a scientifically rigorous sub-x score. I could also have a graph of your best ao5,ao12, ao100, ao1000 over time. Another thing I thought would be interesting is to find a confidence interval for randomly sampling 5 solves and finding a tournament average of 5 and showing you the confidence interval for what score you would get at a tournament based on a solving session.
Does something like this already exist that I could add on to?
I have very little data of my own solves. I would love to have a excel file of someones solves to use when developing this.
I'm curious what kinds of features would be useful to people.
