Question

我在pandas数据框中有几列。基于每一列，我需要创建一个新列。这个功能有效：

def f(row):
    if row['col_1'] == 0:
        val = 'Neutral'
    elif row['col_1'] > 0:
        val = 'Growth'
    else:
        val = 'Contraction'
    return val

df['New_Col_1'] = df.apply(f(row) , axis=1)

但由于我有几列用于比较（col_2，col_3等），我想将列的名称作为参数传递给函数。

def f(row,col_name):
    if row[col_name] == 0:
        val = 'Neutral'
    elif row[col_name] > 0:
        val = 'Growth'
    else:
        val = 'Contraction'
    return val

df['New_Col_1'] = df.apply(f(row,'col_1') , axis=1)

但是，有一个错误。它表示参数'row'未定义。我该如何克服这个问题？

Answer 1

结帐df.loc[]，您可以将两个参数视为行规格和列规格，因此您可以像这样使用它：

df['New_Col_1'] = 'Contraction' # Default, to be overwritten below
df.loc[df['col_1'] == 0, 'New_Col_1'] = 'Neutral'
df.loc[df['col_1'] > 0, 'New_Col_1'] = 'Growth'

Answer 2

您可以使用 df.loc [condition，column_name] = value 来过滤df并写入新值：

df['New_Col_1'] = None # initial
df.loc[df.col1==0, 'New_Col_1'] = 'Neutral'
df.loc[df.col1>0, 'New_Col_1'] = 'Growth'
df.loc[df.col1<0, 'New_Col_1'] = 'Contraction'

Answer 3

df.apply（）缺少评论中提到的lambda函数。

def f(row,col_name):
    if row[col_name] == 0:
        val = 'Neutral'
    elif row[col_name] > 0:
        val = 'Growth'
    else:
        val = 'Contraction'
    return val

df['New_Col_1'] = df.apply(lambda row: f(row,'col_1') , axis=1)

根据使用函数的列满足的条件在pandas中创建新列

3 个答案: