- Joined
- Oct 13, 2016
- Messages
- 15
- Likes
- 7
- Degree
- 0
So I'm starting off with the domcop expiring data, training a regression model to predict price(as listed in the price column) considering as many of the available metrics as possible. The set of metrics domcop provides is actually really, really good but I may enrich it with some other things too. The basic idea is to be able to find potentially undervalued expiring domains, as well as predict prices for all the expired domains(all listed as $8).
The thing is I'm not sure the listed price is the best metric to train with. In my ideal world, I'd actually like to be able to get the prices of actual domain sales and use that to enhance the metric data in the domcop archive, then train on that. The point here is that I know there are plenty of domains that have been listed at a price but just didn't sell.
If anyone can give me any pointers or wants to partner up I'd be happy to work with you. I don't have a lot of money right now to actually invest in domains. This is really just my first practical application of machine learning in the real world.
I'm probably going to be sharing results here and looking for feedback on how the predictions turn out, so fel free to snatch any gems that come up. If you have better data sources I might want to try out either post them here or PM if you're interested in exploring this more on the down low.
The thing is I'm not sure the listed price is the best metric to train with. In my ideal world, I'd actually like to be able to get the prices of actual domain sales and use that to enhance the metric data in the domcop archive, then train on that. The point here is that I know there are plenty of domains that have been listed at a price but just didn't sell.
If anyone can give me any pointers or wants to partner up I'd be happy to work with you. I don't have a lot of money right now to actually invest in domains. This is really just my first practical application of machine learning in the real world.
I'm probably going to be sharing results here and looking for feedback on how the predictions turn out, so fel free to snatch any gems that come up. If you have better data sources I might want to try out either post them here or PM if you're interested in exploring this more on the down low.