In my journey to explore the vast world of Machine Learning and Deep Learning, I delved into various time series prediction models.
The primary aim was to understand, implement, and evaluate different models on their capability to forecast time series data, specifically stock prices. The Close
price was predominantly chosen as a feature for model building.
- Data contain 'timestamp', 'Open', 'High', 'Low', 'Close', 'Volume' dataframes.
- Split the data into training and test sets.
- Scaled the data for smoother model ingestion.
- Conducted parameter tuning using grid search to identify the optimal changepoint and seasonality prior scales.
- Utilized the best performing parameters for predictions on the test data.
- Employed random search over a broad parameter space for hyperparameter optimization.
- Assessed the model's performance on a hold-out test set.
- Implemented these classic time series forecasting methods.
- Benchmarked their performance using metrics like MAE (Mean Absolute Error) and RMSE (Root Mean Squared Error).
- Utilized dates as a single feature to predict the 'Close' prices.
- Analyzed its forecasting power with MAE and RMSE.
- 1 Day Data (40 Data):
- 12 Hours Data (90 Data):
- 1 Hour Data (1000 Data):
The code and methods highlighted above offer a simplified representation of the entire process. The insights gleaned from the results were both fascinating and invaluable in understanding stock price behaviors.
Connect with me on LinkedIn.
For more insights into my work, check out my latest project: tafou.io.
I'm always eager to learn, share, and collaborate. If you have experiences, insights, or thoughts about RL, Prophet, XGBoost, SARIMA, ARIMA, or even simple Linear Regression in the domain of forecasting, please create an issue, drop a comment, or even better, submit a PR!
Let's learn and grow together! 🌱