Spaces:
Sleeping
Sleeping
File size: 2,743 Bytes
87e984d 3562719 87e984d 928f9ee |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 |
---
title: DataSynthis_ML_JobTask
emoji: 📈
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.49.0
app_file: app.py
pinned: false
license: mit
allow_internet: true
---
# Stock Price Forecasting App
This application uses three different models (ARIMA, Prophet, and LSTM) to forecast stock prices.
## ================================================================================
## FINAL RECOMMENDATIONS
## ================================================================================
Based on the comprehensive evaluation:
1. BEST PERFORMING MODEL: LSTM
- Lowest RMSE: $5.39
2. KEY FINDINGS:
- ARIMA Model:
* Simpler and faster to train
* Better for short-term forecasts
* Assumes linear relationships
* RMSE: $28.98
* MAPE: 11.57%
- Prophet Model:
* Excellent at capturing seasonality and trends
* Handles missing data and outliers well
* Provides uncertainty intervals
* RMSE: $16.29
* MAPE: 6.97%
- LSTM Model:
* Captures non-linear patterns
* Better for complex time series
* Requires more data and computation
* RMSE: $5.39
* MAPE: 2.06%
3. RECOMMENDATIONS:
- For production deployment, consider ensemble methods combining all three models
- Prophet is excellent for interpretability and trend analysis
- LSTM performs well when sufficient training data is available
- ARIMA provides quick baseline forecasts
- Regularly retrain models with new data
- Monitor prediction intervals and confidence bounds
- Consider external factors (news, market sentiment) for better predictions
4. MODEL SELECTION GUIDE:
- Use ARIMA for: Quick forecasts, baseline comparisons, stationary data
- Use Prophet for: Seasonal patterns, interpretable results, business forecasts
- Use LSTM for: Complex patterns, non-linear relationships, large datasets
5. LIMITATIONS:
- Stock prices are inherently unpredictable
- Past performance doesn't guarantee future results
- Models should be used as decision support tools, not sole decision makers
- Consider risk management and diversification strategies
- All models assume patterns will continue into the future
## Features
- Real-time stock data fetching from Yahoo Finance
- Multiple forecasting models
- Interactive visualizations
- Customizable forecast periods
## Models
1. **ARIMA** - Traditional statistical model
2. **Prophet** - Facebook's time series forecasting
3. **LSTM** - Deep learning neural network
## Usage
1. Enter a stock ticker symbol (e.g., AAPL, GOOGL)
2. Select forecast period (1-90 days)
3. Choose which model(s) to use
4. Click "Generate Forecast"
⚠️ **Disclaimer**: For educational purposes only. Not financial advice. |