200 points by datascienceninja 6 months ago flag hide 17 comments
john_tech 6 months ago next
Interesting project! Which machine learning algorithms did you use?
john_tech 6 months ago next
I used a combination of Random Forest and SVM algorithms for this project. How about you?
john_tech 6 months ago next
We used a combination of technical analysis indicators and fundamental data. Some examples include moving averages, RSI, and P/E ratio.
ml_engineer 6 months ago prev next
Nice work! Have you considered using LSTM for time series prediction in stock market?
ml_engineer 6 months ago next
Wow that's a great idea! Yes, I'll definitely give it a try. Thanks for the suggestion!
data_scientist 6 months ago prev next
What's your data preprocessing pipeline look like?
bs_detector 6 months ago prev next
The stock market can't be predicted with 100% accuracy with machine learning algorithms. Can you explain how you avoid overfitting?
creator 6 months ago next
Absolutely, I used a combination of cross-validation and feature selection to avoid overfitting. I also made sure to train the model on multiple years of data and evaluated its performance on more recent data.
ml_researcher 6 months ago prev next
I agree with your point about overfitting. Have you tried regularization techniques, such as L1 or L2 regularization?
creator 6 months ago next
Yes, I used L2 regularization, which helped improve the model's generalization ability.
algorithm_expert 6 months ago prev next
What kinds of features do you feed your model? I'm curious about your feature engineering process.
open_source_advocate 6 months ago prev next
Have you considered releasing your code as open source?
creator 6 months ago next
That's a great idea! I'll definitely consider it.
finance_expert 6 months ago prev next
How do you handle the non-stationarity of financial time series data in your model?
creator 6 months ago next
We used a combination of feature scaling and normalization techniques to handle the non-stationarity of the data. Additionally, we used a rolling window approach to train the model on more recent data, which helped improve its performance.
technical_analyst 6 months ago prev next
Nice work! What metrics did you use to evaluate the model's performance?
creator 6 months ago next
We used a combination of metrics, such as accuracy, precision, recall, and F1 score. Additionally, we evaluated the model's performance on a holdout set to ensure its generalization ability.