Question

这是我的示例代码

dataset_current=dataset_seq['Motor_Current_Average']
dataset_consistency=dataset_seq['Consistency_Average']

#technique with non-overlapping the values(for current)
dataset_slide=dataset_current.tolist()
from window_slider import Slider
import numpy
list = numpy.array(dataset_slide)
bucket_size = 336
overlap_count = 0
slider = Slider(bucket_size,overlap_count)
slider.fit(list)      
empty_dictionary = {}
count = 0
while True:
  count += 1
  window_data = slider.slide()
  empty_dictionary['df_current%s'%count] = window_data
  empty_dictionary['df_current%s'%count] =pd.DataFrame(empty_dictionary['df_current%s'%count])
  empty_dictionary['df_current%s'%count]= empty_dictionary['df_current%s'%count].rename(columns={0: 'Motor_Current_Average'})
  if slider.reached_end_of_list(): break
  locals().update(empty_dictionary)


#technique with non-overlapping the values(for consistency)
dataset_slide_consistency=dataset_consistency.tolist()
list = numpy.array(dataset_slide_consistency)
slider_consistency = Slider(bucket_size,overlap_count)
slider_consistency.fit(list)      
empty_dictionary_consistency = {}
count_consistency = 0
while True:
  count_consistency += 1
  window_data_consistency = slider_consistency.slide()
  empty_dictionary_consistency['df_consistency%s'%count_consistency] = window_data_consistency
  empty_dictionary_consistency['df_consistency%s'%count_consistency] =pd.DataFrame(empty_dictionary_consistency['df_consistency%s'%count_consistency])
  empty_dictionary_consistency['df_consistency%s'%count_consistency]= empty_dictionary_consistency['df_consistency%s'%count_consistency].rename(columns={0: 'Consistency_Average'})
  if slider_consistency.reached_end_of_list(): break
  locals().update(empty_dictionary_consistency)
import pandas as pd
output_current ={}
increment = 0
while True:
   increment +=1
   output_current['dataframe%s'%increment] = pd.concat([empty_dictionary_consistency['df_consistency%s'%count_consistency],empty_dictionary['df_current%s'%count]],axis=1)

output_current ={}
length = len(empty_dictionary_consistency)
increment = 0
i = 0
while i < length:
  increment +=1
  output_current['windows%s'%increment] = pd.concat([empty_dictionary_Timestamp['df_Timestamp%s'%increment],empty_dictionary_current['df_current%s'%increment],empty_dictionary_flowrate['df_flowrate%s'%increment],empty_dictionary_level['df_level%s'%increment],empty_dictionary_consistency['df_consistency%s'%increment],empty_dictionary_label['df_label%s'%increment]],axis=1)
  locals().update(output_current)

我的问题是我有26296行的数据集我们拥有的标签列的数量为 24662个案例，显示为0，表示机器正在运行 1633例，表示1表示（机械停止）注意：标签是根据机器停止运行或运行的日期创建的，我们已经为数据集添加了标签。

如何使用上述不平衡数据集获得auc_roc得分。有24662个案例显示0，只有1633个案例显示1达到auc_roc评分至少为85％？另外，我们如何使用这种滑动窗口技术训练任何分类器？通过此操作，我们能够从26296行的给定数据集中创建79个窗口

如何将分类器应用于滑动窗口技术？

0 个答案: