|
|
|
|
|
"""
|
|
|
|
|
|
models包功能：
|
|
|
|
|
|
该包主要实现了4种主要的时间序列预测模型：向量自回归(VAR)、自回归移动平均(ARIMA)、
|
|
|
|
|
|
季节性自回归移动平均 (SARIMA) 以及随机森林（Random Forest）。
|
|
|
|
|
|
|
|
|
|
|
|
modules:
|
|
|
|
|
|
- utils.py: 包含用于数据读取和数据保存的通用函数。
|
|
|
|
|
|
- VAR_model.py: 实现向量自回归 (VAR)预测模型。
|
|
|
|
|
|
- ARIMA_model.py: 实现自回归移动平均 (ARIMA)预测模型。
|
|
|
|
|
|
- SARIMA_model.py: 实现季节性自回归移动平均 (SARIMA)预测模型。
|
|
|
|
|
|
- RF_model.py: 实现随机森林（Random Forest）预测模型。
|
|
|
|
|
|
|
|
|
|
|
|
函数：
|
|
|
|
|
|
- run(): 是整个预测过程的主函数。该函数首先读取数据，然后调用所有模型进行预测，
|
|
|
|
|
|
最后把预测结果保存到文件中。
|
|
|
|
|
|
|
|
|
|
|
|
使用示例:
|
|
|
|
|
|
from models import run
|
|
|
|
|
|
run_result = run(forecast_target=target_column, exog_columns=exog_columns_list, models=[VAR_Forecasting])
|
|
|
|
|
|
"""
|
|
|
|
|
|
from typing import List, Type
|
|
|
|
|
|
from . import utils, VAR_Forecasting, ARIMA_Forecasting, SARIMA_Forecasting, RF_Forecasting
|
|
|
|
|
|
|
|
|
|
|
|
__all__ = [
|
|
|
|
|
|
'utils', 'VAR_Forecasting', 'ARIMA_Forecasting', 'SARIMA_Forecasting',
|
|
|
|
|
|
'RF_Forecasting'
|
|
|
|
|
|
]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def run(
|
|
|
|
|
|
forecast_target: str,
|
|
|
|
|
|
exog_columns: List[str],
|
|
|
|
|
|
steps: int = 20,
|
|
|
|
|
|
file_path: str = 'data/normalized_df.csv',
|
|
|
|
|
|
models: List[Type] = [
|
|
|
|
|
|
VAR_Forecasting, ARIMA_Forecasting, SARIMA_Forecasting, RF_Forecasting
|
|
|
|
|
|
],
|
|
|
|
|
|
) -> None:
|
|
|
|
|
|
"""
|
|
|
|
|
|
执行数据读取、预处理、模型训练、预测并保存预测结果等一系列步骤的主函数。
|
|
|
|
|
|
|
|
|
|
|
|
Args:
|
|
|
|
|
|
forecast_target (str): 需要被预测的目标变量的列名。
|
|
|
|
|
|
exog_columns (List[str]): 用于预测的特征变量的列名列表。
|
|
|
|
|
|
steps (int, default=20): 需要进行预测的步数。
|
|
|
|
|
|
file_path (str, default='data/normalized_df.csv'): 数据文件的路径。
|
|
|
|
|
|
models (List[Type]) : 需要运行的预测模型列表，默认包括VAR、ARIMA、SARIMA和Random Forest模型。
|
|
|
|
|
|
|
|
|
|
|
|
Returns:
|
|
|
|
|
|
None
|
|
|
|
|
|
"""
|
|
|
|
|
|
# 载入数据
|
|
|
|
|
|
input_df = utils.read_csv(file_path)
|
|
|
|
|
|
|
|
|
|
|
|
# 使用每个模型进行预测并保存结果
|
|
|
|
|
|
for model in models:
|
|
|
|
|
|
try:
|
|
|
|
|
|
model_name = model.__name__.split('.')[-1]
|
|
|
|
|
|
print(f"正在执行 {model_name} 模型进行预测...")
|
|
|
|
|
|
|
|
|
|
|
|
# 调用模型进行预测
|
|
|
|
|
|
model_df = model.run(input_df, forecast_target, exog_columns,
|
|
|
|
|
|
steps)
|
|
|
|
|
|
|
|
|
|
|
|
# 保存预测结果
|
|
|
|
|
|
utils.save_csv(model_df, f'data/{model_name}_df.csv')
|
|
|
|
|
|
|
|
|
|
|
|
print(f"{model_name} 模型的预测结果已保存至 data/{model_name}_df.csv")
|
|
|
|
|
|
|
|
|
|
|
|
except Exception as e:
|
|
|
|
|
|
print(f"{model_name} 模型预测过程出现错误: {e}")
|
|
|
|
|
|
print("所有模型预测都已完成。")
|
