<p>用途:</p>
<pre><code>#remove column App, compare and get sum of Trues
a0 = df.drop('App', 1).eq(0).sum()
#a0 = df.set_index('App').eq(0).sum()
#alternative with select only Feature columns
#a0 = df.filter(like='Feature').eq(0).sum()
#alternative with select all columns without first
a0 = df.iloc[:, 1:].eq(0).sum()
print (a0)
Feature1 6
Feature2 7
Feature3 2
Feature4 2
Feature5 4
Feature6 6
Feature7 1
Feature8 7
dtype: int64
</code></pre>
<p>与<code>1</code>类似:</p>
<pre><code>a1 = df.drop('App', 1).eq(1).sum()
#a1 = df.set_index('App').eq(1).sum()
#alternative
#a1 = df.filter(like='Feature').eq(1).sum()
#alternative
a1 = df.iloc[:, 1:].eq(1).sum()
print (a1)
Feature1 2
Feature2 1
Feature3 6
Feature4 6
Feature5 4
Feature6 2
Feature7 7
Feature8 1
dtype: int64
</code></pre>
<p>加上<a href="http://pandas.pydata.org/pandas-docs/stable/generated/pandas.Index.value_counts.html" rel="nofollow noreferrer">^{<cd2>}</a>:</p>
<pre><code>a = df.drop('App', 1).apply(pd.value_counts).T.add_prefix('count_')
print (a)
count_0 count_1
Feature1 6 2
Feature2 7 1
Feature3 2 6
Feature4 2 6
Feature5 4 4
Feature6 6 2
Feature7 1 7
Feature8 7 1
</code></pre>
<p>或与列表理解:</p>
<pre><code>L = [df[x].value_counts() for x in df.columns.difference(['App'])]
a = pd.concat(L, 1).T.add_prefix('count_')
print (a)
count_0 count_1
Feature1 6 2
Feature2 7 1
Feature3 2 6
Feature4 2 6
Feature5 4 4
Feature6 6 2
Feature7 1 7
Feature8 7 1
</code></pre>