New Idea, what do you think

[Influxes]

So a while ago I was thinking of making a service that uses machine learning to appraise accounts. The project failed because I was unable to get a large enough data set to make accurate predictions. I want to try this project again, and I think I can do it. However, to get the data set I would need a lot of people filling out the form I made. Would anyone be willing to help me compile a data set? I think the best possible way to collect the data would be to look at the OG/Semi-OG section of MCM and look for sold accounts. I am looking to make this a free to use service.

If anyone wants to help me build this data set, please fill out this FORM with any accounts you can find.

If you have any suggestions, please tell me. I would like this to be as successful as possible

 

[Influxes]

I know this project would be a big undertaking, but I think this would be a very interesting project. And might be very helpful for fast and accurate appraisals.

 

[Pebbly]

I feel like this would take a lot of time and it would be very hard to actually have the machine to learn how to appraise based on all the different aspects that go into the appraisals of an account, Like in the form, Email Access, Tid, etc. I like the idea though.

 

[Influxes]

I feel like this would take a lot of time and it would be very hard to actually have the machine to learn how to appraise based on all the different aspects that go into the appraisals of an account, Like in the form, Email Access, Tid, etc. I like the idea though.

I tried to make as little variables as possible to see where that got me.

 

[Pebbly]

I tried to make as little variables as possible to see where that got me.

Each account is unique in its on way I feel like, so even if you had little variables playing as the factor, it would still take a lot. If the machine happened to get the price right for say one account, it probably would not get the price right for another account completely different from the one it got right

 

[Influxes]

Each account is unique in its on way I feel like, so even if you had little variables playing as the factor, it would still take a lot

That is true. I was mostly thinking of this being a way to get a quick iffy estimate without having to contact someone.[DOUBLEPOST=1573073380][/DOUBLEPOST]

Not sure what the demand for this would be like, although the idea of using machine learning is a big step. The only thing is if you want to be accurate you would need a lot of data, such as a dictionary (basically) for OG names, then make low letter combinations more pricy depending if they make sense. Then see the use of each word and up the value the more it's used.

A lot goes into this and personally, I don't believe it will be ever accurate enough for the use you're hoping it to get.

That is true. It was mostly just a concept that I thought was really cool. The main problem with this idea is not actually making the machine learning program. I just find the data collection very daunting