基於列數據生成列描述的優雅方法

Question

我有一個pandas數據幀：

index    data1    data2
1        30       20
2        20       10  
3        40       90

我想生成一個描述數組，每行一個，指示數據是否重要。

我定義重要值超過25，所以我想要以下數組：

['data1 was significant', '',  'data1 was significant\ndata2was significant']

我知道我可以遍歷每一行並檢查每一列並構建一個數組，但我想知道是否有一種優雅的方法來使用pandas來做到這一點。

Answer 1

使用dot技巧 ：

df = pd.DataFrame({
    'data1': [30, 20, 40],
    'data2': [20, 10, 90]
}, index=[1,2,3])

df.gt(25).dot(df.columns + ' was significant\n').str.strip().tolist()
# ['data1 was significant', '', 'data1 was significant\ndata2 was significant']

Answer 2

或者，使用np.where 。

[''.join(x) for x in np.where(df > 25, df.columns + ' was significant\n', '')] 
['data1 was significant\n',
 '',
 'data1 was significant\ndata2 was significant\n']

Answer 3

或者，使用apply

In [323]: (df.gt(25).apply(lambda x: '\n'.join(
                       ['%s was significant' % c for c, v in x.iteritems() if v]), axis=1)
             .tolist())
Out[323]: ['data1 was significant', '', 'data1 was significant\ndata2 was significant']