TheProNoob: forum

Hi all,

Google form link here, 

Please make sure that you enter the CPRs that you see in an EMPTY SLOT. The CPRs that you see when someone is in a slot is theirs, it is not yours.

It will take 10-15 mins of your time to extract the data and fill out the form.

tl;dr Minimal reading is provided all in the link above, below is a more wordy explanation of what I'm doing/proposing.

What is this?

I've started an initiative to collect a bunch of data in a centralised manner to help us figure out what the hidden contributions to CPR are. I propose to do this via machine learning regression, simple visual inspection and other more complex techniques, for which I need this data. The more data we have, the more precisely we can extract these relationships.

Ideally, I get a bunch of data over a long period of time of "snapshots" of people's progression through 2.0 (with skill levels past 100). So, fill this out every 2-3 months if you'd like with an updated picture of what your CPRs look like, or whenever it is convenient.

You can choose to provide the data anonymously by inserting a non-related torn ID in the form. If you choose to do this, please remember what unique id you gave me, and fill this form out again with the same ID later on. This is because crime experience (CE) is innate to your account and not visible. Although I can do some treatments for differing levels of CE, it is better that you provide me a unique tag that tags your data as yours.

This dataset will be refreshed whenever there are significant new contributions (i.e. every 50-100 new datapoints, or whenever I have time, whichever comes first), with the first edition being made public in 2-3 months. This dataset will be fully anonymised - there should be no way to trace your stats back to you unless you have abnormally strange stats (e.g. top 10 HoF or superbuster CE).

What do you get out of this?

Ideally, you will get an idea of what crimes/personal stats you need to work on to help your faction out in succeeding OC2.0 at a higher rate.

At worst, smarter/more motivated people than I will get a public dataset to investigate these relationships.

We know from initial observation that crime skill past 100 likely contributes, as well as workstats etc.

If we can draw any conclusions from the data that you provide me, we get to advise players how to specialise in each specific role.

The method/details/dataset sharing (nerd stuff):

CPRs see the most significant variance at roles in higher level OCs. That is why I only request data of these t7/t8 OCs, since it is likely that low-level CPRs give very little information. Likely I will also add t9/t10 CPRs when they are released, as well as other t7/t8 crimes.

I propose to analyse the data with regression techniques - we can use techniques such as random forests and Kolmogorov-Arnold networks with symbolic regression to extract/approximate the relationships that contribute to CPRs. Of course, simple visualisation will be used as well. If you have suggestions here I'm happy to hear them, post below.

The dataset/analysis will be published on an open-source github repo (link will be added soon). This way, other players will be free to continue on this work. I will not miss the first dataset reporting deadline, but I cannot promise anything after that. It will depend on the motivation of the contributors (the people sharing their data) to see if I publish more than a single dataset. I will update this post with a github link once I have set it up. The data will likely be provided in a csv format, unless it gets extremely large (unlikely) at which point it will likely be stored in a different format. 

The analysis, in the form of jupyter notebooks, and the python environment files will be available in the repository.

When can we expect the reports/results?

I'll be blunt, I don't know! Probably whenever I'm not busy and I have enough data to deem it worth my time.

I'm quite busy, but at the very least the data will be published/refreshed on the open-source github repository on a regular basis.

Why am I doing this?

It's fun trying to strip away the mystery behind hidden mechanics!

Any dm without an offer containing this information will be ignored
PM Me:
- Company type
- Company rating (stars)
- Salary

Work Stats:
 Manual labor 200,328
Intelligence 449,181
Endurance 182,561 

Looking for high salary, rig/tv preferred

Sent when I see the event. Attacks will be paid, msgs likely ignored.

Open while the post reads [OPEN]

Closed

Buying 10

Closed now

Valid until post reads [CLOSED], attack and I'll pay when online.

Filtering those that can’t read is a design choice, honestly. 

To address your point, of course it would be better to have full api access to accounts or the equivalent in a script that tracks the relevant data. I haven’t done this because I don’t think it’s necessary. CPRs don’t change dramatically week to week. If someone is willing to make a script for this, that would be just fine too? I’d be happy to incorporate the data. They will still need to manually provide full API access at least once to scrape crime nerve expenditure.

Providing full api access to your account to strangers is also a dangerous thing that many former scripters have been found to abuse. So I think there is a good reason why relying on a trusted middleman is a good idea instead (torn.report)

Part of data science is knowing which parts of your data aren’t useful and cutting it out and throwing it away. Data cleaning is 80% of data science/machine learning and 20% is actually making the models. I will have to do this anyway.

If people can’t be bothered to take the 5 mins to fill out a small form, they simply don’t care enough, and that’s just fine!

Happy to hear the criticism though.

This is optional - you can use your torn id you wish. The option to anonymise themselves is there for people that want to protect their information.

It's now in its own thread here https://www.torn.com/forums.php#/p=threads&f=2&t=16446637&b=0&a=0,

You make a good point about visibility.

Yes - if you can be bothered, it would be valuable to fill it out like once every 3 months or so

Hi all,

Google form link here, 

Please make sure that you enter the CPRs that you see in an EMPTY SLOT. The CPRs that you see when someone is in a slot is theirs, it is not yours.

It will take 10-15 mins of your time to extract the data and fill out the form.

tl;dr Minimal reading is provided all in the link above, below is a more wordy explanation of what I'm doing/proposing.

What is this?

I've started an initiative to collect a bunch of data in a centralised manner to help us figure out what the hidden contributions to CPR are. I propose to do this via machine learning regression, simple visual inspection and other more complex techniques, for which I need this data. The more data we have, the more precisely we can extract these relationships.

Ideally, I get a bunch of data over a long period of time of "snapshots" of people's progression through 2.0 (with skill levels past 100). So, fill this out every 2-3 months if you'd like with an updated picture of what your CPRs look like, or whenever it is convenient.

You can choose to provide the data anonymously by inserting a non-related torn ID in the form. If you choose to do this, please remember what unique id you gave me, and fill this form out again with the same ID later on. This is because crime experience (CE) is innate to your account and not visible. Although I can do some treatments for differing levels of CE, it is better that you provide me a unique tag that tags your data as yours.

This dataset will be refreshed whenever there are significant new contributions (i.e. every 50-100 new datapoints, or whenever I have time, whichever comes first), with the first edition being made public in 2-3 months. This dataset will be fully anonymised - there should be no way to trace your stats back to you unless you have abnormally strange stats (e.g. top 10 HoF or superbuster CE).

What do you get out of this?

Ideally, you will get an idea of what crimes/personal stats you need to work on to help your faction out in succeeding OC2.0 at a higher rate.

At worst, smarter/more motivated people than I will get a public dataset to investigate these relationships.

We know from initial observation that crime skill past 100 likely contributes, as well as workstats etc.

If we can draw any conclusions from the data that you provide me, we get to advise players how to specialise in each specific role.

The method/details/dataset sharing (nerd stuff):

CPRs see the most significant variance at roles in higher level OCs. That is why I only request data of these t7/t8 OCs, since it is likely that low-level CPRs give very little information. Likely I will also add t9/t10 CPRs when they are released, as well as other t7/t8 crimes.

I propose to analyse the data with regression techniques - we can use techniques such as random forests and Kolmogorov-Arnold networks with symbolic regression to extract/approximate the relationships that contribute to CPRs. Of course, simple visualisation will be used as well. If you have suggestions here I'm happy to hear them, post below.

The dataset/analysis will be published on an open-source github repo (link will be added soon). This way, other players will be free to continue on this work. I will not miss the first dataset reporting deadline, but I cannot promise anything after that. It will depend on the motivation of the contributors (the people sharing their data) to see if I publish more than a single dataset. I will update this post with a github link once I have set it up. The data will likely be provided in a csv format, unless it gets extremely large (unlikely) at which point it will likely be stored in a different format. 

The analysis, in the form of jupyter notebooks, and the python environment files will be available in the repository.

When can we expect the reports/results?

I'll be blunt, I don't know! Probably whenever I'm not busy and I have enough data to deem it worth my time.

I'm quite busy, but at the very least the data will be published/refreshed on the open-source github repository on a regular basis.

Why am I doing this?

It's fun trying to strip away the mystery behind hidden mechanics!

I've an initiative to collect a bunch of data to help us figure out what the hidden contributions to CPR are (via machine learning regression) - can do so anonymously (but please remember what unique id you gave me, I need people to fill out the form with data over the course of a few months)

Google form link here

We'd like it if Crime EXP (the ranks) remained accessible through the API, or make it visible in game too, even after migration to OC2.0, please.

a

B

b

b

You're in a rush to renew your donator status, browse the bazaars, buy the wrong box.

B - Evolution looking for 1 more for our 25/25 PA attempts :)

B - Evolution looking for 1 more for our 25/25 PA attempts :)

B - Evolution looking for 1 more for our 25/25 PA attempts :)