Talk:Composter

Could we also rank them by % chance, not just Alphabetically?
I.e., like the Snapshot description Page:

10% chance: all saplings, all leaves, all types of seeds (except nether wart), (sea) grass, sweet berries, and (dried) kelp. 20% chance: dried kelp blocks, tall grass, sugar cane, cactus, vines, and melon slices. 50% chance: wheat, lily pads, melons, sea pickles, pumpkins, apples, potatoes, beetroots, carrots, cocoa beans, all mushrooms, mushroom stem blocks (not mushroom blocks), all ferns, and all (tall) flowers. 80% chance: hay blocks, bread, cookies, and baked potatoes. 100% chance: cake and pumpkin pie.

It also helps show those 5 only seemingly specific % amounts. Easier to keep track of same-#'s. Yilante 1 /16 /19  5:02 p.m. 76.209.248.192 01:02, 17 January 2019 (UTC)


 * Th table can be sorted by the chance to compost, by clicking the column header. Having a compact per-chance list in addition to the table might be a good idea, though. 「 ディノ 奴 千？！ 」☎ Dinoguy1000 01:32, 17 January 2019 (UTC)


 * I've added so hopefully it might sort correctly now. – Nixinova Nixinova sig image 1.png Nixinova sig image 2.png 01:35, 17 January 2019 (UTC)


 * And it looks very niice. But as pointed-out in the MC Forum, "their % 's are weird."


 * So we should probably also analyze the best-ranges of given plants /crops /Crafted-Food, yields. So for example, do each by basically-efficiency of input-type to amount of output for it.


 * To specifically quote, "As things stand you're always better off composting with raw materials." ("Shadow" icon).  Yilante 1 /17 /19  2:33 a.m. 76.209.248.192 10:33, 17 January 2019 (UTC)

Percentages
I really don't think the percentage for wheat seeds is 10%. After extensive testing, I'm convinced it's 12.5%, or one chance in 8 rather than 10. I can do a full-blown chi-square test if anyone wants, but you can see for yourself - run a full chest of wheat seeds through a composter, and you'll get ~31 bone meal.

I haven't yet tested the other items in the "10%", but it seems likely that they also have the higher percentage. TheShadowLC (talk) 13:41, 18 January 2019 (UTC)

I just realized that I made an error in my calculations. A hypothesis test for the true percentage being greater than 10% on my current dataset gives a p-value of 0.137 - a tad high, but nowhere near high enough to reject the null hypothesis. Solution: Collect more data! TheShadowLC (talk) 14:30, 18 January 2019 (UTC)

Sigh, a tad *low*, I meant to type. TheShadowLC (talk) 14:40, 18 January 2019 (UTC)

Okay. Having run 3 double chests of wheat seeds (10368 total) through a composter, I got 196 bone meal, corresponding to 1372 composter levels. I calculate the following:

A one-sided hypothesis of whether the true mean proportion is greater than 0.1 has a p-value of 1.3e-22. I therefore reject the null hypothesis that the true mean proportion is 0.1.

I calculate a 99% confidence interval of [0.12375, 0.14090] for the true mean proportion. That's a range of 1 in 7.0971 to 1 in 8.0803.

The percentage could conceivably be higher than 12.5%. Doing a hypothesis test on that gives a p-value of 0.0138 - much higher than before, but still rather low. TheShadowLC (talk) 18:25, 18 January 2019 (UTC)


 * I just checked looking at several different items that can be composted. The percentages for 4 different items in the 10% category averaged to 11.3%. None of the items I tested were at or below 10%, and none exceeded 12.5%. I also tested 3 items each from the 20%, 50%, and the 80% categories. All recorded percentages were slightly above expected numbers across different items averaging 22.4%, 54.0%, and 82.3%. I agree with ShadowLC that either the real odds are different then the ones listed in the snapshot info or it occurred to me that it's possible some other system might be twisting the odds and giving an increased percentage and at other times may give a lower percentage. The percentages I found were only 1-4% higher then expected values. It might simply be they've configured the composter to always give those percentages at minimum. I used 3 doublechests of items for each item tested as well. Amefeu (talk) 01:23, 24 January 2019 (UTC)