So,
Essentially I have data, odds etc on just about every greyhound that has ever run.
I can't code, I was thinking maybe head to fiverr and use a computer scientist to create a tool to create power ratings and deduce odds.
What exactly is it I should be asking / looking for ? Model learning ? Prediction model? Data analysis ? Were talking a huge amount of data... theres a lot of analysis / AI related terms I dont really understand / know the difference between.
Best way to model odds?
Didn't you say you had ChatGPT access ? If you don't have V4 , with advanced data analysis, then it's worth the £20 a month sub. Upload some sample data, ask the right questions, tell it what you want to do, and take it from there 
It will save you a lot of work in the long run. Oh, and get Python installed, you'll need it - it will even tell you how. You don't even need coding knowledge
It will save you a lot of work in the long run. Oh, and get Python installed, you'll need it - it will even tell you how. You don't even need coding knowledge
Yeah I have successfully had GPT 4 predicting for me. The problem is that with its message memory limit, it begins to stray and makes a mess. Even when I send it chat history in a file it still seems to end up changing its methods for the worse.sniffer66 wrote: ↑Sat Mar 02, 2024 11:03 pmDidn't you say you had ChatGPT access ? If you don't have V4 , with advanced data analysis, then it's worth the £20 a month sub. Upload some sample data, ask the right questions, tell it what you want to do, and take it from there
It will save you a lot of work in the long run. Oh, and get Python installed, you'll need it - it will even tell you how. You don't even need coding knowledge
Also, I can't input the huge data files I have. I'm a little limited to getting it to predict odds off a small sample of data which is much less accurate.
I've found the best thing to do is upload a sample, then get it to give you the python code to run locally. I had that issue with the size of the environment per user. If you get issues with the code, just paste the errors back in and it will fix them. Takes a bit of to and fro but you get there.
I've got sheets with 100k rows and the above works fine. If it can get your headers and a few rows as a sample, it's usually pretty good
I've got sheets with 100k rows and the above works fine. If it can get your headers and a few rows as a sample, it's usually pretty good
sniffer66 wrote: ↑Sat Mar 02, 2024 11:58 pmI've found the best thing to do is upload a sample, then get it to give you the python code to run locally. I had that issue with the size of the environment per user. If you get issues with the code, just paste the errors back in and it will fix them. Takes a bit of to and fro but you get there.
I've got sheets with 100k rows and the above works fine. If it can get your headers and a few rows as a sample, it's usually pretty good
Will give this a go tomorrow. Had been thinking I probably needed it to write some python code so it would stop changing its weightings / prediction model without telling me
