The Thunder's Place PE data "study"
I had a little free time a couple of days ago and decided to finally process the data from the PE Statistics site here on Thunder’s. It seemed like an easy task but it actually took a lot of time and I didn’t even manage to do everything I meant to do at the beginning. I did however calculate the total erect gains of 2294 members (pun intended) and the flaccid gains of 1067. I wanted to calculate gains for various intervals of time, such as 3,6,12,24 months, the top 10 gainers in each category (that’s not at all a particularly hard thing to do, I just won’t do it now), etc.
In order to process the data I’ve downloaded it in it’s .csv form and copied all the info to a simple text file. After that I used a Python script (that’s what I did most of the time) to filter the data so it doesn’t mess up my results.
[irrelevant]I have to admit, I could've done that with no classes (no object oriented programming) and in 50 or so lines, but I took a more structured approach because I've forgotten Python a bit and don't feel confident enough to write some cryptic, magical lines of code, and, to be honest - it seemed like more fun. It turned out a bit of a mess though, because I've really forgotten how to write in Python and I'll not be sharing my code with you guys, because I don't like it myself. And most of you probably wont care. :D [irrelevant]
Anyway, on to a more important matter - how I filtered my results. I removed:
- All entries of all users that weren’t in any sensible limits in order to remove “fake” ones (mostly ones added by mistake, not purposely delusive ones, part of an evil conspiracy). I decided I’d use 2.25-12 for EL, 2.25-10 for EG, 2.25-10 for FL, 2.25-8 for FG and 0.5-3 for Erect Width. There was no statistical analysis prior to choosing those limits, such as mean +- 2*standard deviation, they were chosen based purely on common sense. All entries outside of those limits were, as I said, removed. If you guys decide that the intervals should be different, I'll have no problem changing them and evaluating the entries again, but I doubt that the impact will be big, I'd guess no more than 0.5% of the people would be outside the intervals I've chosen and thus the results won't be greatly influenced.
- All users with one or zero valid entries for obvious reasons - the gain cannot be calculated with only one valid entry.
- I was planning to remove all entries that were very far away from the average of all the entries for a given member, but that turned out to be inconsistent, unnecessary and quite frankly - a bit tiresome to program.
After that I calculated gains from the first entry/measurement to the last one and made some “charts” with the help of the R statistical programming language. The distribution wasn’t unexpected but interesting nonetheless. There were a couple of dozen of members who seemed to have a negative gain (just an example in my defence). I haven’t removed those results as they could possibly be true and are statistically unimportant (as in - the numbers for the other gains remain unchanged). I also calculated some curious stats just for fun, such as total length gain.
I do not claim that the results are perfectly correct, after all the whole “thing” depends on the truthfulness of the stats, people have entered. There were a lot of members with one or more “fake” entries and a lot of members with only one entry, but I think the filtering system worked well. It “threw out” more than half of the members, but 2 thousand is still an impressive number. Also, I had an idea about processing all member signatures, but I realized that even html parsing and regular expressions can't help me, because of the tons of different ways people format their “stats signatures”. A lot of members don't update the PE data regularly but keep their sigs updated, so that could result in a way more accurate and detailed “research”, but I fear it's nearly impossible. So that's what you get.
Before I continue I’d like to add that you can ask me for whatever statistics you want, I’ll do my best to make you happy. :D
1. Erect Length Gain:
- Average gain - 0.68 inches. Have in mind that this is calculated when taking the minuses and zeros into account. It’s ~0.77 without those members who have 0.0 or negative gains.
- Maximal gain - 3.9 inches.
- Histogram (a graphical representation of the data, similar to plots, charts, etc.) - elghist.jpg attachment.
- Total erect length gained by the members processed - 1552.32 inches.
2. Erect Girth Gain:
- Average gain - 0.3 inches. Again, this is with negatives and zeroes included.
- Maximal gain - 3.0 inches. But it’s the gain of this member and it seems pretty fake. The maximal realistic EG gain is 2.375 inches.
- Histogram - egghist.jpg attachment.
- Total erect girth gained by the members processed - 700.59 inches.
3. Erect Width Gain:
- Average gain - 0.1 inches.
- Maximal gain - 0.95 inches. But, again, it’s Craven’s so the real maximal gain would be 0.756 inches.
- Histogram - ewghist.jpg attachment.
- Total erect width gained by the members processed - 222.93 inches.
4. Erect Volume Gain:
- Average gain - 3.2 cubic inches.
- Maximal gain - 30.31 cubic inches and 26.46 taking the second place. In both cases the increase in volume is over 250%. That could be an interesting statistic to make too.
- Histogram - evghist.jpg attachment.
- Total erect volume gained by the members processed - an astonishing cubic 7326.48 inches.
5. Flaccid Length Gain:
- Average gain - 0.64 inches.
- Maximal gain - 4.5 inches. I’m having some difficulty finding the user with that gain due to the crappy realization of the flaccid filtering (it works, but I can’t search users easily) and I honestly don’t want to work on it now, so it may or may not be fake. The second place is with a gain of 3.81 which is quite plausible (not that 4.5 is not for FL).
- Histogram - flghist.jpg attachment.
- Total flaccid length gained by the members processed - 687.32 inches. Notice that the gains are way lower than the EL ones, but the reason behind that (except the obvious one) is that, as I said before, “only” around 1000 members have entered their flaccid statistics.
6. Flaccid Girth Gain:
- Average gain - 0.37 inches.
- Maximal gain - 3.293.
- Histogram - fgghist.jpg attachment.
- Total flaccid girth gained by the members processed - 394.07 inches.
I feel like I’m forgetting something… Oh, well, I guess I can post again if I remember it.
As I said, if you want any statistics feel free to ask. :)
RE-Started (01.10.2012): 4.75'' NBPEL, 4.33'' EG (overall), 3'' NBPFL --- Now: 5.1'' NBPEL, 4.75'' EG (overall), 3.5'' NBPFL
Short-term goal: 5.5'' NBPEL, 4.5'' EG (overall), 3.5'' NBPFL --- Long term goal: 7.5'' NBPEL, 5.5'' EG (overall), 5'' NBPFL
Wish me luck! :)