如何根据某些条件减去 Pandas 数据框的行？

Question

我正在对 this 数据集进行分析。

使用下面的代码后，我得到了数据的清理版本。

covid_df.drop(columns = ["Sno", "Time"], inplace = True)

covid_df["State/UnionTerritory"] = covid_df["State/UnionTerritory"].replace({
    "Bihar****": "Bihar",
    "Maharashtra***": "Maharashtra", 
    "Madhya Pradesh***": "Madhya Pradesh", 
    "Karanataka": "Karnataka",
    "Telangana": "Telengana",
    "Himanchal Pradesh": "Himachal Pradesh",
    "Dadra and Nagar Haveli": "Dadra and Nagar Haveli and Daman and Diu",
    "Daman & Diu": "Dadra and Nagar Haveli and Daman and Diu"
    })

invalid_states = ["Cases being reassigned to states", "Unassigned"]

for invalid_state in invalid_states:
  invalid_state_index = covid_df.loc[covid_df["State/UnionTerritory"] == invalid_state, :].index
  covid_df.drop(index = invalid_state_index, inplace = True)

covid_df = covid_df.groupby(["State/UnionTerritory", "Date"], as_index = False).sum()

covid_df["Date"] = pd.to_datetime(covid_df["Date"])
covid_df.sort_values(by = ["State/UnionTerritory", "Date"], inplace = True)

此清理后的数据具有每个 State/UnionTerritory 每个日期的累积案例。如何提取每个 State/UnionTerritory 的每日新病例？

这是我试过的：

daily_cases_data = [list(covid_df.iloc[0, 2:])]

for index in range(1, covid_df.shape[0]):
  previous_row = covid_df.iloc[index - 1, :]
  current_row = covid_df.iloc[index, :]

  if previous_row["State/UnionTerritory"] == current_row["State/UnionTerritory"]:
    daily_cases_data.append(list(current_row[2:] - previous_row[2:]))
  else:
    daily_cases_data.append(list(current_row[2:]))

有没有更有效的方法？

Answer 1

编辑后的答案：使用groupby.shift 如此处所示：

df = pd.DataFrame(
    {
        'state': ['a', 'a', 'a', 'b', 'b', 'c', 'c', 'c', 'c'],
        'cumul': [1, 2, 5, 3, 4, 5, 8, 9, 9]
    }
)
df['quantity'] = df['cumul'] - df.groupby('state')['cumul'].shift()

上一个回答：

您可以使用 shift。例如：

df = pd.DataFrame({'cumul': [0, 2, 3, 5, 7]})
df['quantity'] = df - df.shift(1)

quantity 将是：

   quantity
0    NaN
1    2.0
2    1.0
3    2.0
4    2.0

然后您可以 fillna 或仅将数量中的第零个值更改为累积中的第零个值。

编辑：首先应用您的条件来准备数据框:-)

Answer 2

回答我自己的问题感觉很奇怪，但这对我有用

grouped_df = covid_df.groupby("State/UnionTerritory")
daily_cases_df = pd.DataFrame()

for state in covid_df["State/UnionTerritory"].unique():
  group = grouped_df.get_group(state)
  cases_group = group.iloc[:, 2:] - group.shift(1).iloc[:, 2:]
  cases_group.iloc[0, :] = group.iloc[0, 2:]
  group = pd.concat([group.iloc[:, :2], cases_group], axis = 1)
  
  daily_cases_df = pd.concat([daily_cases_df, group])

如何根据某些条件减去 Pandas 数据框的行？

How to subtract rows of a Pandas dataframe based upon some conditions?

python

pandas

dataframe

pandas-groupby