View unanswered posts | View active topics It is currently Fri Jan 20, 2017 8:56 am



Welcome
Welcome to XP-Waste!

You are currently viewing our boards as a guest, which gives you limited access to view most discussions and access our other features. By joining our free community, you will have access to post topics, communicate privately with other members (PM), respond to polls, upload content, and access many other special features. In addition, registered members also see less advertisements. Registration is fast, simple, and absolutely free, so please, join our community today!


Reply to topic  [ 46 posts ]  Go to page Previous  1, 2, 3
 Farming Research 
Author Message
Researcher
Researcher

Joined: Sat Apr 16, 2011 9:40 am
Posts: 1216
RSN: Meredith
Post Re: Farming Research
archermanme wrote:
Is there any value to logging data by patch. In terms of yield deaths and seeds returned per patch? Vs Just logging a whole run. I get that it would give data per patch but would that be worthwhile getting?

There shouldn't be any difference between logging per patch vs per run, except that if one specific patch were anomalous for whatever reason, we would be able to catch it if we logged per patch.

_________________


Tue Oct 30, 2012 3:26 pm
Profile
Researcher
Researcher
User avatar

Joined: Sat Jun 18, 2011 5:54 am
Posts: 1038
Location: Wherever I may roam
RSN: metta
Post Re: Farming Research
The only patch that it would make sense to have a higher death rate is the port phasmatus patch because mortania is full of plague and death, but I highly doubt there is a difference.

_________________
Image

Image

Image

Image

Witch King Obtaurian, Dark Lord of the Sith wrote:
I am better than Aero Dreamz


Tue Oct 30, 2012 3:47 pm
Profile
Researcher
Researcher

Joined: Sat Apr 16, 2011 9:40 am
Posts: 1216
RSN: Meredith
Post Re: Farming Research
I've heard rumors that the Falador patch yields more than others, but I don't think that holds any truth. Regardless, I think it's better to play it safe in this case, since it doesn't take that much more effort.

_________________


Tue Oct 30, 2012 4:44 pm
Profile
Researcher
Researcher
User avatar

Joined: Sat Jun 18, 2011 5:54 am
Posts: 1038
Location: Wherever I may roam
RSN: metta
Post Re: Farming Research
Mod ash posted some insightful things about farming on the old school forums. I did not get to read it myself but I have found posts made by other people about it which get the idea across.

This is what the guy posted, as we later see it fits with our data very well so I think he correctly understood mod ash
Quote:
Herbs, like allotments and hops, work as follows:

You start with 3 lives.
Each time you get an item, you roll the dice.
If you rolled unsuccessfully, you lose a life. When you've lost all 3 lives, the patch becomes empty.
So if you roll unsuccessfully every time, you'd still get 3 items.
In theory, if you rolled successfully every time, you'd get an infinite number of items.

If you use compost, you start with 4 lives instead.
If you use supercompost, you start with 5 lives instead.

Secateurs make you more likely to roll successfully, so you're statistically more likely to get more items, but not guaranteed.

Anyway, thought you might like to know.

Just a bit of information I found, just sharing.


I will derive an equation for the probability of receiving n herbs from a patch (excluding juju and greenfingers)
In the quote the guy talks about losing lives. Suppose that when you pick a herb the chance of losing a life is p. I'll just examine the case when you have 5 lives because we almost always use supercompost.
In order to get n herbs you must lose 4 lives by the (n-1)th attempt and then on the nth attempt you must lose a life

Probability of losing 4 lives by the (n-1)th attempt is Image

Probability of losing 4 lives by the (n-1)th attempt and then on the nth attempt lose another life= Image

Using the definition of nCr this becomes
Image
So that is the formula for the probability for receiving n herbs from 1 patch (excluding juju and greenfingers)

Zigzagpaul noted the number of torstol he harvested for 2099 patches. Using trial and error I found that the value of p which minimises the weighted square difference between zigzagpaul's findings and the expected findings using the formula was 64.74%. The chance of the plant "losing a life" after each harvest is 64.74% This fits the data beautifully. In the chart below the blue is zigzagpaul's data and the red is what we'd expect for p=64.74%
Image

Note that p=64.74% is unique to torstols at level 99. Mod ash said that herbs/allotments/hops harvest depends on the level requirement for that plant and your current farming level. We expect that it barely matters for herbs.

_________________
Image

Image

Image

Image

Witch King Obtaurian, Dark Lord of the Sith wrote:
I am better than Aero Dreamz


Mon Aug 12, 2013 9:58 pm
Profile
Tech Admin
User avatar

Joined: Thu Jul 28, 2011 12:32 am
Posts: 1652
RSN: Calisme
Post Re: Farming Research
That's actually really cool, and makes a lot of sense. Thanks for this, Thai.

_________________
Image


Mon Aug 12, 2013 11:42 pm
Profile
Researcher
Researcher
User avatar

Joined: Sat Jun 18, 2011 5:54 am
Posts: 1038
Location: Wherever I may roam
RSN: metta
Post Re: Farming Research
Thai guy wrote:
Image
So that is the formula for the probability for receiving n herbs from 1 patch (excluding juju and greenfingers)

I realised that this is a Negative binomial distribution with r=5 and p=chance of not losing a life, the only difference is that in farming when you "fail" and lose a life you still get a herb whereas the way a negative binomial distribution is defined when you "fail" you wouldn't get a herb. Because of this you get 5 free herbs, it is easy for adjust for this by just setting n equal to n+5 in any formulae for the negative binomial distribution. This could be helpful in future.
http://en.wikipedia.org/wiki/Negative_b ... stribution

The distribution of herbs yielded when using juju and an aura is difficult to model. This is because there are many degrees of freedom: number of yield attempts, number of juju activations, number of aura activations meaning that an expression for the probability of getting a set number of herbs will have 3 summations in it and they are hard to evaluate without specified values of number of herbs gained. The process of getting herbs with juju and greenfingers is a multinomial process. Modelling the number of herbs gained from 5 patches is even more confusing and it will have 19 degrees of freedom. But we needed to somehow find the distributions of herbs from a whole farm run in order to extract useful information from farming logs which keep track of total herbs from a farm run.
Firstly I wrote a script to create a table of the probability of getting n herbs from y yields with specified probabilities of juju and greenfingers activating.
The formula for the chance of getting n herbs from y yields is
Image
Where x1, x2, x3 are the number of times you get 1, 2, 3 herbs from a yield attempt and k1, k2 k3 are the chance of getting 1, 2, 3 herbs from a yield attempt.

Having got that table I was then able to find the probability of getting n herbs without assuming the number of yield attempts. The number of yield attempts is given by the first formula in this post.
The probability of getting n herbs from 1 patch is then
Image
Which I used my table to evaluate. (max is just some upper limit which it is extremely unlikely to get above)

The probability distribution is approximately a normal distribution so I will make the assumption that it is. I used the table to find the mean and standard deviation of the distribution. The number of herbs gained from 5 patches is the sum of 5 normal distributions and it will itself be a normal distibution with mean 5 times higher and standard deviation sqrt(5) times higher.

My table was set up so that I could enter the chance of losing a life and it would return the mean and standard deviation.
I call the number of herbs from 5 patches N
With the normal distribution approximation the best estimation of the probability of getting N herbs is Image There is a function for this in all spreadsheet softwares. Phi(N) is the normal distribution function for the number of herbs from 5 patches. Phi(N) changes as I change the chance of losing a life figure.

I grouped Justin and Donald's farming log into specific herbs for different aura tiers. 5 groups were large enough to analyze. Torstol at T4, Snapdragons at T0, T3 and T4, and Dwarf weed at T4. Justin and Donald's farming log is dated, using runetracker Justin and I got the average level they were at in each group. I compared the observed frequency to the frequency predicted by the normal distribution approximation. With trial and error I found the value of "chance of losing a life" which minimised the weighted mean squared error of the predicted probability of getting N herbs.

With the 5 points from Junstin and Donalds log, and the 1 point from Zigzagpaul's log I got an approximate equation relating player farming level and the herb's farming requirement. It was:
Chance of losing a life= 1.2672-0.008516*(farming level)+0.003404*(farming requirement)
However, this was not accurate for my data on guams so the trend which Jagex used must be quite nonlinear. I do think though that we can get a reasonable estimation for herb yield in high level herbs at high farming levels.

The 6 figures for chance of losing a life are here https://docs.google.com/spreadsheet/ccc ... xckE#gid=0

_________________
Image

Image

Image

Image

Witch King Obtaurian, Dark Lord of the Sith wrote:
I am better than Aero Dreamz


Sat Aug 17, 2013 9:40 am
Profile
Display posts from previous:  Sort by  
Reply to topic   [ 46 posts ]  Go to page Previous  1, 2, 3

Who is online

Users browsing this forum: No registered users and 2 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
suspicion-preferred