Ingress 中，MUFG 储存胶囊输出道具的规律是怎样的？
MUFG Capsule - final report
This analysis is based on over 4 months worth of data (1299 item replications observed in total).
Most of this is already known and was reported before, but I want to do one final overview with latest numbers, because I'm ending my experiment.
1) The observed combined replication probability is 0.949% (95% CI is 0.899-1.001%).
2) The probability is universal among all item types.
3) Replication rates for all item types are more or less compatible with 1% probability (see the attached chart). The only exception is XMP8, but the deviation is not very significant, only 2.8 standard deviations below prediction - I invoke the Rule of Nice Numbers™ and blame it on a bad luck ;-)
4) The time between two replications is random, averaging at 23.0 hours and it seems normally distributed with sigma around 1-2 hours (see the attached histogram).
5) I was curious about the possibility that NIA has changed the probability at some point, so I plotted the attached graph of how replication rate evolved in time (30 days rolling average). Well, the results are inconclusive - it seems like in summer the probability was higher and than it dropped, but the error bars are much higher at the beginning, since I did not have as many capsules as later.
6) I did one more independent test to confirm the general idea of this theory - the attached table contains observed and predicted probability for a capsule with 95 items to produce 0, 1, 2, 3, 4 and 5 items respectively. The prediction is based on 1% replication probability for every item. The observation reasonably fits the prediction.
https://pic3.zhimg.com/b93dab0b54a10ce9091c5624fa28a8d2_r.jpg
https://pic4.zhimg.com/d6a5ea8253de9d142ed65ec71b7d3dc3_r.jpg
頁:
[1]