文章预览
承接Part I： 增强深度学习模型评估以进行股市预测--Part I 训练和测试模型 我们使用自定义验证指标训练TFT模型。 步骤1：训练过程 from neuralforecast.models import TFT from neuralforecast import NeuralForecast from neuralforecast.losses.pytorch import HuberMQLoss from pandas.tseries.offsets import CustomBusinessDay from pytorch_lightning.loggers.tensorboard import TensorBoardLogger from pytorch_lightning.callbacks import ModelCheckpoint, EarlyStopping from tensorboard.backend.event_processing import event_accumulator import os import numpy as np def create_custom_trading_days (start_date: str, end_date: str, market: str = "NYSE" ) -> CustomBusinessDay: market_dates = obtain_market_dates(start_date, end_date, market) trading_days = pd.DatetimeIndex(market_dates.index) all_dates = pd.date_range(start=start_date, end=end_date, freq= 'B' )
………………………………
