Data based Neural networks accuracy is poor

#1

Hi Friends - I’m wondering how to go about creating neural networks for data based problems. I was trying my hands on with a house price prediction problem. I tried a simple, as well as few layered networks(took example from a kaggle competition kernel), but the training accuracy never gets past 1%. So I think am doing something terribly wrong. I would like to get some suggestions on how to approach these kind of problems.

Here is the github link for the notebook and data. - https://github.com/iosnewbie2016/housepricepredict

Can someone pls guide me on this. Help is much appreciated.

Thanks.

0 Likes

(Jeremy Howard (Admin)) #2

Try watching video 4 of the new course and following the approach there.

0 Likes

#3

Thanks Jeremy.

0 Likes

(ecdrid) #4

@MLNewbie
I don’t think so…
Here’s my and I am happy with my position on LB…

Let me know if you ran into some trouble on this nbs
(hadn’t made much changes to the reference notebooks though)

1 Like

(Brintia) #5

I just saw this free webinar on Nueral networks and thought it might be useful :slight_smile: https://www.brintia.com/webinar-the-keys-to-understanding-neural-networks/

0 Likes