如何将来自在Python循环中传递的函数中的变量附加到数据框？

huangapple
go评论98阅读模式
英文:

How do I append variables to a dataframe coming from a function passed in a Python loop?

问题

Here is the translated code section you provided:

  1. 我的目标是获得这个输出：

campaign | mean | left | right
A | 5% | 0% | 10%
B | 8% | 2% | 6% etc.

  2. 目前，我正在使用一个循环内的Python函数来获取这4个变量：
  3. campaign、mean、left、right。以下是该函数：
  5. ```python
  6. import numpy as np
  7. def bootstrap_ci(df, variable, classes, repetitions = 1000, alpha = 0.05, random_state=None):
  8.     df = df[[variable, classes]]
  9.     bootstrap_sample_size = len(df)
  11.     mean_diffs = []
  13.     for i in range(repetitions):
  14.         bootstrap_sample = df.sample(n = bootstrap_sample_size, replace = True, random_state = random_state)
  15.         mean_diff = bootstrap_sample.groupby(classes).mean().iloc[1,0] - bootstrap_sample.groupby(classes).mean().iloc[0,0]
  16.         mean_diffs.append(mean_diff)
  18.     left = np.percentile(mean_diffs, alpha/2*100)*(-1)
  19.     right = np.percentile(mean_diffs, 100-alpha/2*100)*(-1)
  20.     mean = -(df.groupby(classes).mean().iloc[1,0] - df.groupby(classes).mean().iloc[0,0])

接下来，我正在创建一个循环来运行所有我的广告系列的自举函数（这个工作）。

  1. for i in campaigns:     
  2.     print(Lift for: {i}')     
  3.     bootstrap_ci(df[df['campaign_name']==i],'conversion','group')

然后，我尝试将以下代码添加到循环函数中，以将left、right、mean这些变量附加到df（这个不起作用）：

  1. df = pd.DataFrame([]) --初始化一个新的df
  3. for i in campaigns:     
  4.     print(Lift for: {i}')     
  5.     bootstrap_ci(df[df['campaign_name']==i],'conversion','group')
  6.     df = df.append(i, mean, left, right) --将相关变量附加到df

请注意，你在附加变量时需要将它们包装在字典或Series中，以确保正确附加到DataFrame中。你可以使用类似以下的代码：

  1. df = pd.DataFrame([]) # 初始化一个新的df
  3. for i in campaigns:     
  4.     print(f'Lift for: {i}')     
  5.     result = bootstrap_ci(df[df['campaign_name']==i],'conversion','group')
  6.     df = df.append({'campaign': i, 'mean': result['mean'], 'left': result['left'], 'right': result['right']}, ignore_index=True)

这样，你将能够将每个广告系列的结果添加到DataFrame中。

英文:

My goal is to have this output:

  1. campaign | mean | left | right 
  2. A        | 5%   | 0%   | 10% 
  3. B        | 8%   | 2%   | 6% etc.

Right now, I am using a python function within a loop to get the 4 variables:
campaign, mean, left, right. Here is the function:

  1. import numpy as np
  2. def bootstrap_ci(df, variable, classes, repetitions = 1000, alpha = 0.05, random_state=None):df = df[[variable, classes]]bootstrap_sample_size = len(df)
  4. mean_diffs = []
  6. for i in range(repetitions):
  7.     bootstrap_sample = df.sample(n = bootstrap_sample_size, replace = True, random_state = random_state)
  8.     mean_diff = bootstrap_sample.groupby(classes).mean().iloc[1,0] - bootstrap_sample.groupby(classes).mean().iloc[0,0]
  9.     mean_diffs.append(mean_diff)
  11.     left = np.percentile(mean_diffs, alpha/2*100)*(-1)
  12.     right = np.percentile(mean_diffs, 100-alpha/2*100)*(-1)
  13.     mean = -(df.groupby(classes).mean().iloc[1,0] - df.groupby(classes).mean().iloc[0,0])

Next, I am creating a loop to run the bootstrap function for all my campaigns (this works).

  1. for i in campaigns:     
  2.     print(Lift for: {i}')     
  3.     bootstrap_ci(df[df['campaign_name']==i],'conversion','group')

Then, I tried adding this code to the loop function to append the variables left, right, mean (this does not work):

  1. df = pd.DataFrame([]) --initiate a new df
  3. for i in campaigns:     
  4.     print(Lift for: {i}')     
  5.     bootstrap_ci(df[df['campaign_name']==i],'conversion','group')
  6.     df = df.append(i, mean, left, right) --append the relevant variables to df

Please advise as the df returns nothing...

答案1

得分: 0

Append new items with df.at[index, column] = value, as follows:

  1. import numpy as np
  2. import pandas as pd
  4. def bootstrap_ci(df, variable, classes, repetitions=1000, alpha=0.05, random_state=None):
  5.     df = df[[variable, classes]]
  6.     bootstrap_sample_size = len(df)
  7.     mean_diffs = []
  8.     
  9.     for i in range(repetitions):
  10.         bootstrap_sample = df.sample(n=bootstrap_sample_size, replace=True,
  11.                                      random_state=random_state)
  12.         mean_diff = bootstrap_sample.groupby(classes).mean().iloc[1, 0] - bootstrap_sample.groupby(classes).mean().iloc[0, 0]
  13.         mean_diffs.append(mean_diff)
  14.     
  15.     left = np.percentile(mean_diffs, alpha / 2 * 100) * (-1)
  16.     right = np.percentile(mean_diffs, 100 - alpha / 2 * 100) * (-1)
  17.     mean = -(df.groupby(classes).mean().iloc[1, 0] - df.groupby(classes).mean().iloc[0, 0])
  18.    
  19.    # added return statement
  20.    return dict(mean=mean, left=left, right=right)
  23. # initiate a new df
  24. new_df = pd.DataFrame({'mean': [], 'left': [], 'right': []})
  26. for i in campaigns:     
  27.     print('Lift for: {i}')     
  28.     values = bootstrap_ci(df[df['campaign_name'] == i], 'conversion', 'group')
  29.     
  30.     # append the relevant variables to df
  31.     for k, v in values.items():
  32.         new_df.at[i, k] = v
英文:

Append new items with df.at[index, column] = value, as follows:

  1. import numpy as np
  2. import pandas as pd
  4. def bootstrap_ci(df, variable, classes, repetitions=1000, alpha=0.05, random_state=None):
  5.     df = df[[variable, classes]]
  6.     bootstrap_sample_size = len(df)
  7.     mean_diffs = []
  8.     
  9.     for i in range(repetitions):
  10.         bootstrap_sample = df.sample(n=bootstrap_sample_size, replace=True,
  11.                                      random_state=random_state)
  12.         mean_diff = bootstrap_sample.groupby(classes).mean().iloc[1, 0] - bootstrap_sample.groupby(classes).mean().iloc[0, 0]
  13.         mean_diffs.append(mean_diff)
  14.     
  15.     left = np.percentile(mean_diffs, alpha / 2 * 100) * (-1)
  16.     right = np.percentile(mean_diffs, 100 - alpha / 2 * 100) * (-1)
  17.     mean = -(df.groupby(classes).mean().iloc[1, 0] - df.groupby(classes).mean().iloc[0, 0])
  18.    
  19.    # added return statement
  20.    return dict(mean=mean, left=left, right=right)
  23. # initiate a new df
  24. new_df = pd.DataFrame({'mean': [], 'left': [], 'right': []})
  26. for i in campaigns:     
  27.     print('Lift for: {i}')     
  28.     values = bootstrap_ci(df[df['campaign_name'] == i], 'conversion', 'group')
  29.     
  30.     # append the relevant variables to df
  31.     for k, v in values.items():
  32.         new_df.at[i, k] = v

huangapple
  • 本文由 发表于 2023年4月20日 02:00:15
  • 转载请务必保留本文链接：https://go.coder-hub.com/76057568.html
  • pandas
  • python
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定