Question

我正在尝试将此r代码转换为pandas代码。

Dependents.Numeric = recode(start.DF$Dependents, '3+'='3') %>% as.character() %>% as.numeric()
start.DF = start.DF %>% mutate(HouseholdSize = ifelse(is.na(Dependents.Numeric) | is.na(Married), NA,
                                                      ifelse(CoapplicantIncome > 0 | Married == "Yes",
                                                             Dependents.Numeric + 2,
                                                             Dependents.Numeric +1)))
start.DF = start.DF %>% mutate(IncomePC = TotalIncome/HouseholdSize)

我想找到问题的答案并找到答案，但是它不在python中，而是用r编写。因此，如果有人可以帮助我，我将不胜感激。谢谢

Answer 1

R代码：

start.DF <- data.frame(
  Dependents = c('2', '3+', '1'),
  Married = c('yes', NA, 'no'),
  CoapplicantIncome = c(45, 0, 75),
  TotalIncome = c(100, 67, 80)
)
library(dplyr)
Dependents.Numeric = recode(start.DF$Dependents, '3+'='3') %>% as.character() %>% as.numeric()
start.DF = start.DF %>% mutate(HouseholdSize = ifelse(is.na(Dependents.Numeric) | is.na(Married), NA,
                                                      ifelse(CoapplicantIncome > 0 | Married == "Yes",
                                                             Dependents.Numeric + 2,
                                                             Dependents.Numeric +1)))
start.DF = start.DF %>% mutate(IncomePC = TotalIncome/HouseholdSize)
start.DF

  Dependents Married CoapplicantIncome TotalIncome HouseholdSize IncomePC
1          2     yes                45         100             4 25.00000
2         3+    <NA>                 0          67            NA       NA
3          1      no                75          80             3 26.66667

Python代码：

import pandas as pd
import numpy as np
start_df = pd.DataFrame.from_dict({
    'Dependents': ('2', '3+', '1'),
    'Married': ('yes', np.nan, 'no'),
    'CoapplicantIncome': (45, 0, 75),
    'TotalIncome': (100, 67, 80)
})

Dependents_Numeric = start_df['Dependents'].str.replace('3\+', '3').astype(float)
start_df['HouseholdSize'] = np.nan
cond1 = pd.isna(Dependents_Numeric) | pd.isna(start_df['Married'])
cond2 = (start_df['CoapplicantIncome'] > 0) | (start_df['Married'] == 'yes')
start_df.loc[~cond1 & cond2, 'HouseholdSize'] = Dependents_Numeric[~cond1 & cond2] + 2
start_df.loc[~cond1 & ~cond2, 'HouseholdSize'] = Dependents_Numeric[~cond1 & ~cond2] + 1
start_df['IncomePC'] = start_df['TotalIncome'] / start_df['HouseholdSize']
print(start_df)

   CoapplicantIncome Dependents Married  TotalIncome  HouseholdSize   IncomePC
0                 45          2     yes          100            4.0  25.000000
1                  0         3+     NaN           67            NaN        NaN
2                 75          1      no           80            3.0  26.666667

Answer 2

您可以使用 datar 将 R 代码翻译成 python：

>>> from datar.all import (
...   NA, c, f, tibble, as_numeric, recode, if_else, is_na, mutate
... )
>>> 
>>> start_DF = tibble(
...   Dependents = c('2', '3+', '1'),
...   Married = c('Yes', NA, 'No'),
...   CoapplicantIncome = c(45, 0, 75),
...   TotalIncome = c(100, 67, 80)
... )
>>> 
>>> Dependents_Numeric = as_numeric(recode(start_DF.Dependents, **{'3+': '3'}))
>>> 
>>> start_DF >> mutate(
...   HouseholdSize = if_else(
...     is_na(Dependents_Numeric) | is_na(f.Married), 
...     NA,
...     if_else(
...       (f.CoapplicantIncome > 0) | (f.Married == "Yes"),
...       Dependents_Numeric + 2,
...       Dependents_Numeric + 1
...     )
...   ),
...   
...   IncomePC = f.TotalIncome/f.HouseholdSize
... )
>>> # You don't need to write a new mutate. 
>>> # 'HouseholdSize' can be recycled in side the first mutate
>>> # start_DF >> mutate(IncomePC = f.TotalIncome/f.HouseholdSize)
>>>
  Dependents  Married  CoapplicantIncome  TotalIncome  HouseholdSize   IncomePC
    <object> <object>            <int64>      <int64>      <float64>  <float64>
0          2      Yes                 45          100            4.0  25.000000
1         3+      NaN                  0           67            NaN        NaN
2          1       No                 75           80            3.0  26.666667

将R代码转换为python pandas.code

2 个答案: