I’m curious what people think about this. Just a little exercise of curiosity here (and possibly a feature to request)…
I was wondering if there was an easy way to visualize the grading data. Also, I was wondering how well I was in sync with other graders (do I grade harder or easier than everyone else?).
After playing with the data a little, I came up with this…
– I used excel to make buckets of data to make a stacked-column histogram. The idea was to see where most of the gradings were falling.
caveats:
– This is not all the data available – I just used the first 3 pages of gradings. I manually typed it in.
– This type of chart would have to be re-created on the spot, because as books change level over time (due to new gradings), the related gradings for that book would move to another bucket.
– I’m sure this could be made prettier. I just threw it together in excel.
They look nice and provide easier to read data than the grading pages.
If I were to be nitpicky, I’d say there’s no way to tell in the graphic how many people did the grading for each bar. (This is important for weighing the value of the bar)
Cool idea. I agree with @Megumin that the y-axis should be the number of gradings, not a percent. It’s important to know for example if there was only one grading with a level 24 book, but 10 gradings with a level 30 book.
Hm. That’s important data too, but I’m not sure how to put all that data into one chart. There’s value in both the percent and the absolute numbers.
The percent chart allows you to see, apples-to-apples, which way the trend is going.
The absolute numbers allow you to where most people are making comparisons. (Btw, I think it’s interesting that most people – in the dataset I have – are grading the current book as one of their harder books. Or maybe it’s just one of the harder books in the natively library?)
Your attention in the absolutes is first drawn to the center pile of the bell curve, whereas the percents draw your attention to the trend from left to right.
@ForeKred wow thank you for doing all this! I agree doing a histogram would be cool… i’m not sure exactly which graph I like best, maybe the last one?
I think this graph, along with historical grading (how the level changes over time) and potential bounds for the level (Yotsubato historically has oscillated between level 17 & level 19) would be great.
I think all this information could live in a “view more grading information” section. The gradings section right now I think is still the most compelling (people care most about what other people are directly saying and comparing), but agreed this would be helpful and should be in the roadmap.
If you need more data, don’t hesitate to reach out. You don’t have to do everything by hand
Formatting this was a bit of a pain, partly because I didn’t take the time to make the graph axes automatically adjust to different grading levels.
“clumps” - the rockpile charts were not as smooth as I was expecting, but after looking at the data, it made more sense. There’s some books that are a lot more popular than other books and so all the grades associated with those books would put a spike at that level.
It was interesting that the “other book” (the comparison book) tends to be within +/-10 levels of the “primary book”. (ie. Nobody compared an L15 to an L30.)
