Talk:God Catching Alchemy Meister:Gameplay

I'm in the process of collecting data on sales rates, and I intend to share my findings here on the wiki; this page seems the most appropriate place. I'll just post the formulas in the Sales section and provide my data here in talk. (Well, unless someone suggests another place for the formulas.)

At the moment, I've determined that Sekkyaku (I play in jp), times the Raitenritsu or Baihinyoku (I need further evidence to determine which), rounded down, is your base sale rate percentage. All the data aren't in yet, but your Jisseki score seems to add .05 each to Sekkyaku in this.

I've got a script for collecting data; I'm collecting in samples of 50,000 days attempting to sell- each of these samples takes a little over two days to accumulate. I figure my findings will be where I want them in about a week. Any comments: I'll check this Talk page before posting.

--B.J. Black

Sales raw data
So much for "about a week". A bug introduced in my script led me on a wild goose chase and frustrated me into leaving this behind for a few months. But, here is the raw data I have now: (1) This sample, and the next three, are part of the forementioned wild goose chase.

(2) In this sample, I changed the items stocked drastically to gauge its effect (or, as it turned out, lack of effect). It is otherwise identical to the previous sample.

(3) The last sample of the wild goose chase.

For each sample above, my script loaded a save with the numbers shown (only one clerk in the shop) and passed a day in the dungeon, then returned to the workshop to see if there was a sale; 50,000 times. I believe you get one customer per day, minimum, and so Raitenritsu will have no effect until it passes 100%

--B.J. Black

More raw data
I gathered this in the same fashion as above. I expected once Raitenritsu exceeded 100%, more than one customer would come each day. I was still using just one clerk, but this showed that that method was insufficient. I modified my data mining to handle two clerks (each carefully maneuvered into having exactly 100% sale rate) and gathered this (leaving Raitenritsu at 110%): This indicated two things: first, every customer that comes to the shop will select one clerk at random to serve him, and second, the number of customers that come each day is determined without the use of random numbers, so I can skip the 50,000 samples thing- if my clerks sell at 100%, the total number of sales will equal the total number of customers, every time.

Further testing with those clerks showed that the number of customers per day was equal to Raitenritsu times the number of clerks (but, based on my above data, there is a minimum of one customer per day).

Again, if anybody knows the names of these variables in the English translation, feel free to change them.

--B.J. Black

Some more data
Here is the names of the variables in the English translation:

Jisseki = sales (character stat)

Koubaiyoku = Sales (store stat)

Raitenritsu = Attract

Some more data: This indicates two things:

-there is no effect if the sale rate of a clerk exceeds 100%.

-the number of customers is (Attract times the number of clerks times the number of days) rounded down to the nearest whole number

Any comments: I'll check this Talk page before posting.

Link304

Link304 (talk) 22:39, February 4, 2013 (UTC)

Sansei
Well, you found a fact my methodology overlooked entirely. I assumed every day was entirely independent of the prior days, and my data gathering reflects that. Good catch.

I concur with your conclusions. Although there is a chance, depending on your methods, that an alternate explanation for your numbers is true instead- that is, that sales rate over 100% actually can result in multiple sales, and the number of customers is as I originally proposed- the odds of the exact same results for your two clerks is so miniscule I feel it's safe to say you're correct.

Thanks for the translations as well.

--B.J. Black

new test
Thanks for the comments.

I should have give it before (later is better than never...), but here is the methodology I used: I start from the same save file, select my clerk (only 1 clerk in the store), wait N days in town and go back to my workshop.

I agree with your analysis (moreover, using the same save file can affect the random number generator). So I did a new test : I use the same conditions, but with Attract (Raitenritsu) = 205%. I obtain: items sold = 2, 6, 12 for each clerk. This is consistent with the formula. I also did a short test with a sale rate > 200%, but there was no changes. I should test with more data... but I'm too lazy for that... (Gomenasai !!!)

I will now update the topic.

Again, thanks for your comments, and your data collection and analysis is really impressive !

Link304 Link304 (talk) 19:57, February 8, 2013 (UTC)