[RELEASE] Slopey's BPC Market Tool! - New version!

As soon as I see return profits jump from 180k-200k to 300,000cr per return trip I stop the search. Considering there is no way in hell my Lakon 6 with 104 cargo can possibly find a trade run that nets me that in 30 ly from system A to system B ;) I've settled on a 190k run.
I've noticed some weird looking very high profit routes displayed for me and started checking 3 of them. Every single one of those were completely false on selling and buying prices.
Either there was a big market update from FD or people are polluting the database with nonsense (be it with or w/o OCR)...
And those values weren't even old but like 30 minutes ago. :|
I'll keep an eye out for those routes, let's hope those were just 3 for the statistics.
 
It's not that easy - there is massive variation in some commodity prices so it's tricky to do as it would then exclude any valid by low prices which are the exact ones we're looking for to make the best profit.

The BPC *does* filter incoming data, especially where it's above the Average once the commodity is over a certain value. I'll be tightening that up in the future, but for now, the best plan would be for people not to dump data in blindly from EliteOCR, which is not 100% accurate all the time.

I'm going to start flagging up OCR imported data so I can see how large an issue it is, and if it's a significant problem, I may have to remove the import feature until people validate their data better.

I'm also looking to move to registered users in the very near future, and if you consistently upload poor data, you'll get banned from the BPC.

Personally I think EOCR is highly accurate, the recent problems have been due to missing commodity names and a few commodities where only a couple of letters changed getting it confused. Currently in terms of numerical values i've yet to see a single instance of an incorrect value in over 40 scans, so several hundred actual commodities. However I do agree that user error is a big factor and i'm as guilty as others for uploading bogus data as I didn't know commodity names were getting messed up ( i.e. correct values assigned to wrong commodity names). Luckily I found that quickly and have rectified the data.

I agree you should be flagging imported data and perhaps add a property to the csv that requires tools creating that output put there name into it so you can identify the source of the data. I would hate to see you remove the import feature though, because there is no way i'd spend the time to enter this stuff by hand. I do however agree that registered users or perhaps simply unique user ID's would be beneficial.

I would rather unique user Id's (based on hardware perhaps?) as I dislike signing up for stuff as i've got too many accounts already and its clear that very few if any can offer decent security or protection of ones data, even the large corporations. Just my 2cents.
 

Slopey

Volunteer Moderator
Personally I think EOCR is highly accurate, the recent problems have been due to missing commodity names and a few commodities where only a couple of letters changed getting it confused. Currently in terms of numerical values i've yet to see a single instance of an incorrect value in over 40 scans, so several hundred actual commodities. However I do agree that user error is a big factor and i'm as guilty as others for uploading bogus data as I didn't know commodity names were getting messed up ( i.e. correct values assigned to wrong commodity names). Luckily I found that quickly and have rectified the data.

I agree you should be flagging imported data and perhaps add a property to the csv that requires tools creating that output put there name into it so you can identify the source of the data. I would hate to see you remove the import feature though, because there is no way i'd spend the time to enter this stuff by hand. I do however agree that registered users or perhaps simply unique user ID's would be beneficial.

I would rather unique user Id's (based on hardware perhaps?) as I dislike signing up for stuff as i've got too many accounts already and its clear that very few if any can offer decent security or protection of ones data, even the large corporations. Just my 2cents.

I think the biggest problem is that some people are running lower than 1080 or non-standard resolutions, and that's what EliteOCR has issues with. It does seem pretty accurate, but it only takes one digit on a high value commodity and that throws the BPC out by offering a very lucrative trade route which doesn't exist. I'm also seeing some entries where the demand/supply figure is concatenated into the price/sell numbers. I don't ask for demand, so that's coming in from an import most likely - I can't see anyone typing that in on purpose as they'd type something random if they wanted to mess it up.
 
Last edited:
Personally I think EOCR is highly accurate, the recent problems have been due to missing commodity names and a few commodities where only a couple of letters changed getting it confused. Currently in terms of numerical values i've yet to see a single instance of an incorrect value in over 40 scans, so several hundred actual commodities. However I do agree that user error is a big factor and i'm as guilty as others for uploading bogus data as I didn't know commodity names were getting messed up ( i.e. correct values assigned to wrong commodity names). Luckily I found that quickly and have rectified the data.

I agree you should be flagging imported data and perhaps add a property to the csv that requires tools creating that output put there name into it so you can identify the source of the data. I would hate to see you remove the import feature though, because there is no way i'd spend the time to enter this stuff by hand. I do however agree that registered users or perhaps simply unique user ID's would be beneficial.

I would rather unique user Id's (based on hardware perhaps?) as I dislike signing up for stuff as i've got too many accounts already and its clear that very few if any can offer decent security or protection of ones data, even the large corporations. Just my 2cents.

Might be an idea, didn't slopey mention something awhile back about user accounts? Maybe if that was a thing we could have a uploaded by XXX to make it easier to find out where the bad info etc is coming from. I can see you point about the account security though.

My res is at 1080 and so far like you when I check the results it has ll been accurate. Can't say I have done 40 scans though more like 5 to 10 maybe.
 
hi

BPC crash every time i try to upload data.. by hand or with EliteOCR

The station is Hooke City in Khepri system




edit: correct the name system
 
Last edited:

Slopey

Volunteer Moderator
new update not working for me, parsing to 82% then fails. Any thoughts?

I have a fix for this, but as a work-around, turn off Auto-updating while it does the first (very long) update, then turn it back on afterwards. It's long parses (the initial download) which is doing it and the auto-updater tries to do it again (which as it's threaded it can), and it eventually has a record locking issue.

- - - - - Additional Content Posted / Auto Merge - - - - -

hi

BPC crash every time i try to upload data.. by hand or with EliteOCR

The station is Hooke City in Khepri system




edit: correct the name system

I'll look into that one. Thanks :)

Can you post the error message you get please?
 
- - - - - Additional Content Posted / Auto Merge - - - - -



I'll look into that one. Thanks
smile.png


Can you post the error message you get please?



Can't find a log file (?)
But it's simply crash.. no msg..
just windows tell me that the prg have stop working and trying to find a solution.. that's all i have..
 
Last edited:
Error when trying to parse CSV file with new version of ELITEOCR

Ive tried the proper system name AND the proper station name in those fields, and no go.
 
I think the biggest problem is that some people are running lower than 1080 or non-standard resolutions, and that's what EliteOCR has issues with. It does seem pretty accurate, but it only takes one digit on a high value commodity and that throws the BPC out by offering a very lucrative trade route which doesn't exist. I'm also seeing some entries where the demand/supply figure is concatenated into the price/sell numbers. I don't ask for demand, so that's coming in from an import most likely - I can't see anyone typing that in on purpose as they'd type something random if they wanted to mess it up.
I would say every 2 or 3 stations i get a commodity with half number on EliteOCR.
Also keep in mind that most people don't update their EliteOCR and might be using a version where the Indite/Idium get confused.

I personally decided to go out of the popular area and started adding my own stations so at least if an error comes its from me :)
 
Last edited:
Nevermind...


There should be a big line where it says " YOUR FILE NAME MUST BE THE SYSTEM NAME SPELLED CORRECTLY."
 
Last edited:
Still no joy, uninstalled, reinstalled, restarted ed. last install just crashed with the stopped working message as soon as I started it. Such a good tool and nice to have when your'e all alone in the dark. Slopey? helps me!!
 
I think the biggest problem is that some people are running lower than 1080 or non-standard resolutions, and that's what EliteOCR has issues with. It does seem pretty accurate, but it only takes one digit on a high value commodity and that throws the BPC out by offering a very lucrative trade route which doesn't exist. I'm also seeing some entries where the demand/supply figure is concatenated into the price/sell numbers. I don't ask for demand, so that's coming in from an import most likely - I can't see anyone typing that in on purpose as they'd type something random if they wanted to mess it up.

Thats true and it gave me an idea. What about crowd sourcing accuracy of entries? That is users are able to examine a specific stations data and rate it to its accuracy? That way we might be able to spot obvious bogus data caused by users or tools accidentally putting one commodities values into another data or similar? You could even go further and require one validation per user for every 5 uploads of data they make or similar - though that would be somewhat contentious.

Its quite a fascinating problem, how to guarantee or at least minimize errors on human sourced data, solutions to which are probably entire projects in their own right. I do like the idea of crowd sourcing validation of data, seems like a good use of 'assets', but of course it could also be a point of weakness exploited by those who want to mess up the data.
 
Hey Slopey,
I think people are misusing the compatibility between BPC and eliteOCR at the moment. Since release there are lot of people not understanding the implications. It might be more productive to remove the csv import for now. In one or two weeks we could redo it with more acceptabe results and less trolls/idiots. The interest would be lower and the invalid entries would be reduced.
 

Slopey

Volunteer Moderator
Nevermind...


There should be a big line where it says " YOUR FILE NAME MUST BE THE SYSTEM NAME SPELLED CORRECTLY."

That's irrelevant. What needs to be in there is the system name at the start of each CSV line. The BPC will inform you if that's not the case in the next update.

- - - - - Additional Content Posted / Auto Merge - - - - -

Still no joy, uninstalled, reinstalled, restarted ed. last install just crashed with the stopped working message as soon as I started it. Such a good tool and nice to have when your'e all alone in the dark. Slopey? helps me!!

Bit more verbose please - where's it crashing, during parsing? If so, the next update addresses that.
 
Hey Slopey,
I think people are misusing the compatibility between BPC and eliteOCR at the moment. Since release there are lot of people not understanding the implications. It might be more productive to remove the csv import for now. In one or two weeks we could redo it with more acceptabe results and less trolls/idiots. The interest would be lower and the invalid entries would be reduced.

Would it not be possible to remove the OCR ALL function and the also the high confidence ability? That way it forces people to manually approve each commodity? I'd be very sad if it was removed altogether.
 
Back
Top Bottom