股票市场预测
所属分类:其他
开发工具:Python
文件大小:36KB
下载次数:17
上传日期:2020-07-23 19:48:40
上 传 者:
某斌
说明: LSTM 作为预测模型,使用贝叶斯优化算法来实现股票预测的功能
(LSTM as a prediction model, uses Bayesian optimization algorithm to achieve the function of stock forecasting)
文件列表:
AAPL_974.csv (70461, 2020-05-09)
arima.py (2148, 2020-05-09)
gru_bayesian.py (11903, 2020-05-09)
lstm_basic.py (8829, 2020-05-09)
lstm_bayesian.py (11906, 2020-05-09)
该压缩包的网址.txt (50, 2020-07-22)
# Stock-Market-Prediction
This project deals with predicting the opening BID, ASK and PRICE using daily Apple stock data of 4 years. This data would be useful for traders to determine if a stock should be traded as a future, forward, option or some other derivative as they would be able to see the future trend.
## Data Pre-processing and Feature Engineering
1. Check for missing values - no missing values
2. Check if categorical data, encoding needed - no categorical data
3. Check Feature Importance - All features had high significance, so not dropping
4. Split data into train, validation and test in the ratio 60:20:20
5. Feature Scaling - We are normalizing data between zero and one.
## Prediction Models
1. Autoregressive Integrated Moving Average models (ARIMA)
2. Long Short-Term Memory (LSTM)
3. Gated Recurrent Unit (GRU)
## Model Optimization
Bayesian Optimization of GRU and LSTM models.
## Accuracy Measures for Comparison
1. Mean Absolute Error (MAE)
2. Root Mean Squared Error (RMSE)
3. Mean Forecast Error (MFE)
![accuracy comparison3](https://user-images.githubusercontent.com/55213734/81438718-336b7600-913b-11ea-8463-b1b5afcd06ea.PNG)
Even the less complex models like ARIMA can show better performance than the complex neural netwrok models on some datasets.
近期下载者:
相关文件:
收藏者: