复制链接

克隆策略

In [3]:

train_start_date = '2010-01-01'
train_end_date = '2010-10-01'


features = """

ols('residual', mf_net_pct_main_0, mf_net_pct_l_0, 20)

"""


t_ins_m = M.instruments.v2(
    start_date=train_start_date,
    end_date=train_end_date,
    market='CN_STOCK_A',
    instrument_list='',
    max_count=0
)


features_m = M.input_features.v1(features=features)

t_g_features_m = M.general_feature_extractor.v7(
    instruments=t_ins_m.data,
    features=features_m.data,
    start_date='',
    end_date='',
    before_start_days=0
)

t_d_features_m = M.derived_feature_extractor.v3(
    input_data=t_g_features_m.data,
    features=features_m.data,
    date_col='date',
    instrument_col='instrument'
)

[2022-01-31 11:31:16.200351] INFO: moduleinvoker: instruments.v2 开始运行..

[2022-01-31 11:31:18.631505] INFO: moduleinvoker: instruments.v2 运行完成[2.431146s].

[2022-01-31 11:31:18.637740] INFO: moduleinvoker: input_features.v1 开始运行..

[2022-01-31 11:31:18.651700] INFO: moduleinvoker: 命中缓存

[2022-01-31 11:31:18.653899] INFO: moduleinvoker: input_features.v1 运行完成[0.016191s].

[2022-01-31 11:31:18.669297] INFO: moduleinvoker: general_feature_extractor.v7 开始运行..

[2022-01-31 11:31:58.556228] INFO: 基础特征抽取: 年份 2010, 特征行数=315385

[2022-01-31 11:31:58.600178] INFO: 基础特征抽取: 总行数: 315385

[2022-01-31 11:31:58.606566] INFO: moduleinvoker: general_feature_extractor.v7 运行完成[39.937271s].

[2022-01-31 11:31:58.624917] INFO: moduleinvoker: derived_feature_extractor.v3 开始运行..

[2022-01-31 11:32:17.393945] INFO: derived_feature_extractor: 提取失败 ols('residual', mf_net_pct_main_0, mf_net_pct_l_0, 20): Input contains NaN, infinity or a value too large for dtype('float32').

[2022-01-31 11:32:17.939569] INFO: derived_feature_extractor: /y_2010, 315385

[2022-01-31 11:32:18.116409] INFO: moduleinvoker: derived_feature_extractor.v3 运行完成[19.491475s].