PriceForecast/main_yuanyou.py

213 lines
8.9 KiB
Python
Raw Permalink Normal View History

2024-11-08 11:18:40 +08:00
# 读取配置
2024-11-14 16:38:30 +08:00
from config_jingbo import *
2024-11-01 16:38:21 +08:00
from lib.dataread import *
from lib.tools import *
2024-11-22 13:26:10 +08:00
from models.nerulforcastmodels import ex_Model,model_losss,model_losss_juxiting,brent_export_pdf,tansuanli_export_pdf,pp_export_pdf,model_losss_juxiting
2024-11-01 16:38:21 +08:00
import glob
import torch
torch.set_float32_matmul_precision("high")
sqlitedb = SQLiteHandler(db_name)
sqlitedb.connect()
2024-11-01 16:38:21 +08:00
def predict_main():
"""
主预测函数用于从 ETA 获取数据处理数据训练模型并进行预测
参数:
signature (BinanceAPI): Binance API 实例
etadata (EtaReader): ETA 数据读取器实例
is_eta (bool): 是否从 ETA 获取数据
data_set (str): 数据集名称
dataset (str): 数据集路径
add_kdj (bool): 是否添加 KDJ 指标
is_timefurture (bool): 是否添加时间衍生特征
end_time (str): 结束时间
is_edbnamelist (bool): 是否使用 EDB 名称列表
edbnamelist (list): EDB 名称列表
y (str): 预测目标列名
sqlitedb (SQLiteDB): SQLite 数据库实例
is_corr (bool): 是否进行相关性分析
horizon (int): 预测时域
input_size (int): 输入数据大小
train_steps (int): 训练步数
val_check_steps (int): 验证检查步数
early_stop_patience_steps (int): 早停耐心步数
is_debug (bool): 是否调试模式
dataset (str): 数据集名称
is_train (bool): 是否训练模型
is_fivemodels (bool): 是否使用五个模型
val_size (float): 验证集大小
test_size (float): 测试集大小
settings (dict): 模型设置
now (str): 当前时间
etadata (EtaReader): ETA 数据读取器实例
modelsindex (list): 模型索引列表
data (str): 数据类型
is_eta (bool): 是否从 ETA 获取数据
返回:
None
"""
2024-11-01 16:38:21 +08:00
signature = BinanceAPI(APPID, SECRET)
etadata = EtaReader(signature=signature,
classifylisturl=classifylisturl,
classifyidlisturl=classifyidlisturl,
edbcodedataurl=edbcodedataurl,
edbcodelist=edbcodelist,
edbdatapushurl=edbdatapushurl,
edbdeleteurl=edbdeleteurl,
edbbusinessurl=edbbusinessurl
)
# 获取数据
if is_eta:
logger.info('从eta获取数据...')
signature = BinanceAPI(APPID, SECRET)
etadata = EtaReader(signature=signature,
classifylisturl=classifylisturl,
2024-11-01 16:38:21 +08:00
classifyidlisturl=classifyidlisturl,
edbcodedataurl=edbcodedataurl,
edbcodelist=edbcodelist,
edbdatapushurl=edbdatapushurl,
edbdeleteurl=edbdeleteurl,
edbbusinessurl=edbbusinessurl,
2024-11-01 16:38:21 +08:00
)
df_zhibiaoshuju, df_zhibiaoliebiao = etadata.get_eta_api_yuanyou_data(data_set=data_set, dataset=dataset) # 原始数据,未处理
2024-11-01 16:38:21 +08:00
# 数据处理
df = datachuli(df_zhibiaoshuju, df_zhibiaoliebiao, y=y, dataset=dataset, add_kdj=add_kdj, is_timefurture=is_timefurture,
end_time=end_time)
2024-11-01 16:38:21 +08:00
else:
2024-11-14 10:21:25 +08:00
# 读取数据
logger.info('读取本地数据:' + os.path.join(dataset, data_set))
df = getdata(filename=os.path.join(dataset, data_set), y=y, dataset=dataset, add_kdj=add_kdj,
is_timefurture=is_timefurture, end_time=end_time) # 原始数据,未处理
2024-11-01 16:38:21 +08:00
# 更改预测列名称
df.rename(columns={y: 'y'}, inplace=True)
2024-11-01 16:38:21 +08:00
if is_edbnamelist:
df = df[edbnamelist]
df.to_csv(os.path.join(dataset, '指标数据.csv'), index=False)
2024-11-01 16:38:21 +08:00
# 保存最新日期的y值到数据库
# 取第一行数据存储到数据库中
first_row = df[['ds', 'y']].tail(1)
2024-11-01 16:38:21 +08:00
# 将最新真实值保存到数据库
if not sqlitedb.check_table_exists('trueandpredict'):
first_row.to_sql('trueandpredict', sqlitedb.connection, index=False)
2024-11-01 16:38:21 +08:00
else:
for row in first_row.itertuples(index=False):
row_dict = row._asdict()
2024-11-05 15:38:14 +08:00
row_dict['ds'] = row_dict['ds'].strftime('%Y-%m-%d %H:%M:%S')
check_query = sqlitedb.select_data('trueandpredict', where_condition=f"ds = '{row.ds}'")
2024-11-01 16:38:21 +08:00
if len(check_query) > 0:
set_clause = ", ".join([f"{key} = '{value}'" for key, value in row_dict.items()])
sqlitedb.update_data('trueandpredict', set_clause, where_condition=f"ds = '{row.ds}'")
2024-11-01 16:38:21 +08:00
continue
sqlitedb.insert_data('trueandpredict', tuple(row_dict.values()), columns=row_dict.keys())
2024-11-01 16:38:21 +08:00
import datetime
# 判断当前日期是不是周一
2024-11-08 11:18:40 +08:00
is_weekday = datetime.datetime.now().weekday() == 0
2024-11-01 16:38:21 +08:00
if is_weekday:
logger.info('今天是周一,更新预测模型')
2024-11-22 13:26:10 +08:00
# 计算最近60天预测残差最低的模型名称
2024-11-21 14:09:00 +08:00
model_results = sqlitedb.select_data('trueandpredict', order_by="ds DESC", limit="60")
2024-11-22 13:26:10 +08:00
# 删除空值率为40%以上的列
if len(model_results) > 10:
model_results = model_results.dropna(thresh=len(model_results)*0.6,axis=1)
# 删除空行
2024-11-08 11:18:40 +08:00
model_results = model_results.dropna()
modelnames = model_results.columns.to_list()[2:]
2024-11-08 11:18:40 +08:00
for col in model_results[modelnames].select_dtypes(include=['object']).columns:
model_results[col] = model_results[col].astype(np.float32)
# 计算每个预测值与真实值之间的偏差率
for model in modelnames:
model_results[f'{model}_abs_error_rate'] = abs(model_results['y'] - model_results[model]) / model_results['y']
# 获取每行对应的最小偏差率值
min_abs_error_rate_values = model_results.apply(lambda row: row[[f'{model}_abs_error_rate' for model in modelnames]].min(), axis=1)
# 获取每行对应的最小偏差率值对应的列名
min_abs_error_rate_column_name = model_results.apply(lambda row: row[[f'{model}_abs_error_rate' for model in modelnames]].idxmin(), axis=1)
# 将列名索引转换为列名
min_abs_error_rate_column_name = min_abs_error_rate_column_name.map(lambda x: x.split('_')[0])
# 取出现次数最多的模型名称
most_common_model = min_abs_error_rate_column_name.value_counts().idxmax()
2024-11-21 14:09:00 +08:00
logger.info(f"最近60天预测残差最低的模型名称{most_common_model}")
2024-11-01 16:38:21 +08:00
# 保存结果到数据库
if not sqlitedb.check_table_exists('most_model'):
sqlitedb.create_table('most_model', columns="ds datetime, most_common_model TEXT")
sqlitedb.insert_data('most_model', (datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S'), most_common_model,), columns=('ds', 'most_common_model',))
2024-11-01 16:38:21 +08:00
if is_corr:
df = corr_feature(df=df)
df1 = df.copy() # 备份一下后面特征筛选完之后加入ds y 列用
logger.info(f"开始训练模型...")
row, col = df.shape
2024-11-01 16:38:21 +08:00
now = datetime.datetime.now().strftime('%Y%m%d%H%M%S')
ex_Model(df,
horizon=horizon,
input_size=input_size,
train_steps=train_steps,
val_check_steps=val_check_steps,
early_stop_patience_steps=early_stop_patience_steps,
is_debug=is_debug,
dataset=dataset,
is_train=is_train,
is_fivemodels=is_fivemodels,
val_size=val_size,
test_size=test_size,
settings=settings,
now=now,
etadata=etadata,
modelsindex=modelsindex,
data=data,
is_eta=is_eta,
)
2024-11-01 16:38:21 +08:00
2024-11-07 10:45:51 +08:00
logger.info('模型训练完成')
2024-11-07 10:45:51 +08:00
logger.info('训练数据绘图ing')
model_results3 = model_losss(sqlitedb)
# model_results3 = model_losss_juxiting(sqlitedb)
2024-11-07 10:45:51 +08:00
logger.info('训练数据绘图end')
2024-11-01 16:38:21 +08:00
# 模型报告
2024-11-07 10:45:51 +08:00
logger.info('制作报告ing')
title = f'{settings}--{now}-预测报告' # 报告标题
brent_export_pdf(dataset=dataset,num_models = 5 if is_fivemodels else 22,time=end_time,
reportname=reportname,sqlitedb=sqlitedb),
2024-11-12 18:26:20 +08:00
2024-11-07 10:45:51 +08:00
logger.info('制作报告end')
logger.info('模型训练完成')
2024-11-01 16:38:21 +08:00
# # LSTM 单变量模型
# ex_Lstm(df,input_seq_len=input_size,output_seq_len=horizon,is_debug=is_debug,dataset=dataset)
# # lstm 多变量模型
# ex_Lstm_M(df,n_days=input_size,out_days=horizon,is_debug=is_debug,datasetpath=dataset)
# # GRU 模型
# # ex_GRU(df)
# 发送邮件
m = SendMail(
username=username,
passwd=passwd,
recv=recv,
title=title,
content=content,
file=max(glob.glob(os.path.join(dataset,'*.pdf')), key=os.path.getctime),
ssl=ssl,
)
2024-11-21 14:09:00 +08:00
# m.send_mail()
2024-11-01 16:38:21 +08:00
if __name__ == '__main__':
predict_main()