Pipline的recursive_predict的IndexError: list index out of range bug

[suntao2015005848]

在使用Pipline训练NBEATSModel模型后： {"model_type": "forecasting", "ancestor_classname_set": ["NBEATSModel", "PaddleBaseModelImpl", "PaddleBaseModel", "BaseModel", "Trainable", "ABC", "object"], "modulename": "paddlets.models.forecasting.dl.nbeats", "size": {"in_chunk_len": 288, "out_chunk_len": 24, "skip_chunk_len": 0}, "input_data": {"target_dim": 1, "known_cov_dim": 3}} 递归预测测试代码：

#模型直接预测
import pandas as pd

import numpy as np
import matplotlib.pyplot as plt

import datetime
import paddlets
from paddlets import TSDataset
from paddlets import TimeSeries
from paddlets.models.forecasting.dl import * #引入了全部预测模型
from paddlets.models.forecasting import * #引入了全部预测模型
from paddlets.transform import Fill, StandardScaler
from paddlets.metrics import MSE, MAE
import warnings
warnings.filterwarnings('ignore')
import os 
from paddlets.transform import TimeFeatureGenerator
from paddlets.pipeline.pipeline import Pipeline

df = pd.read_csv('/home/aistudio/ydl/fh_power_data.csv')
df = df.filter(items=['monitorTime', 'presentValue'])
target_cov_dataset = TSDataset.load_from_dataframe(
    df,
    time_col='monitorTime',
    target_cols='presentValue',
    known_cov_cols ='monitorTime',
    freq='5min',
    fill_missing_dates=True,
    fillna_method='pre'
)    

# 是否是工作日
time_feature_generator = TimeFeatureGenerator(feature_cols=['is_workday','weekday','hour'])
target_cov_dataset_1 = time_feature_generator.fit_transform(target_cov_dataset)

df_cov = target_cov_dataset_1.to_dataframe()

# 转为df后重新处理为tsdataset
target_cov_dataset_2 = TSDataset.load_from_dataframe(
    df_cov,
    time_col='monitorTime',
    target_cols='presentValue',
    known_cov_cols =['is_workday','weekday','hour'],
    freq='5min',
    fill_missing_dates=True,
    fillna_method='pre'
)    

loaded_model = Pipeline.load("/home/aistudio/ydl/model_nbeats/in288_out24/")
print(len(loaded_model._transform_list))
da1 ,da2 = target_cov_dataset_2.split('2023-09-08 23:55:00')
da3 ,da4 = da2.split('2023-09-09 23:55:00')
print('预测数据：---------------------------------------------------------->')
print(da3.get_target())
print(da3.get_known_cov())

# res = mod.predict(da1)
loaded_model.recursive_predict(da3, predict_length=24)

输出结果报错：

0
预测数据：---------------------------------------------------------->
                     presentValue
monitorTime                      
2023-09-09 00:00:00           6.2
2023-09-09 00:05:00           6.4
2023-09-09 00:10:00           6.2
2023-09-09 00:15:00           7.0
2023-09-09 00:20:00           7.0
...                           ...
2023-09-09 23:35:00           6.3
2023-09-09 23:40:00           6.1
2023-09-09 23:45:00           6.4
2023-09-09 23:50:00           6.9
2023-09-09 23:55:00           6.2

[288 rows x 1 columns]
                     is_workday  weekday  hour
monitorTime                                   
2023-09-09 00:00:00         0.0        5     0
2023-09-09 00:05:00         0.0        5     0
2023-09-09 00:10:00         0.0        5     0
2023-09-09 00:15:00         0.0        5     0
2023-09-09 00:20:00         0.0        5     0
...                         ...      ...   ...
2023-09-13 17:20:00         1.0        2    17
2023-09-13 17:25:00         1.0        2    17
2023-09-13 17:30:00         1.0        2    17
2023-09-13 17:35:00         1.0        2    17
2023-09-13 17:40:00         1.0        2    17

[1365 rows x 3 columns]

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
/tmp/ipykernel_13721/4072872998.py in <module>
     58 
     59 # res = mod.predict(da1)
---> 60 loaded_model.recursive_predict(da3, predict_length=24)
     61 

/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddlets/pipeline/pipeline.py in recursive_predict(self, tsdataset, predict_length)
    280             TSDataset: Predicted results.
    281         """
--> 282         return self._recursive_predict(tsdataset, predict_length)
    283 
    284     def recursive_predict_proba(

/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddlets/pipeline/pipeline.py in _recursive_predict(self, tsdataset, predict_length, need_proba)
    404             pre_data = tsdataset_copy
    405         data_pre_transformed, data_pre_transformed_caches = self.transform(tsdataset=pre_data,
--> 406                                                                            cache_transform_steps=True)
    407 
    408         results = []

/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddlets/pipeline/pipeline.py in transform(self, tsdataset, inplace, cache_transform_steps, previous_caches)
    143         transform_caches = [None] * tansform_list_length
    144         # the first transformer's cache is the origin data
--> 145         if cache_transform_steps and self._transform_list[0].need_previous_data:
    146             transform_caches[0] = tsdataset_transformed
    147 

IndexError: list index out of range

[Sunting78]

您好 可以提供下具体的环境，我们进行复现

[Sunting78]

长时间未回复，该issue已关闭，如仍有问题可以reopen或新开issue。