<h2>设置</h2>
<pre><code>dates = pd.date_range('2016/10/01', '2018/08/16', freq='M')
matrixcols = list('ABCD')
df = pd.DataFrame(np.ones((len(dates), len(matrixcols)), int), dates, matrixcols)
A B C D
2016-10-31 1 1 1 1
2016-11-30 1 1 1 1
2016-12-31 1 1 1 1
2017-01-31 1 1 1 1
2017-02-28 1 1 1 1
2017-03-31 1 1 1 1
2017-04-30 1 1 1 1
2017-05-31 1 1 1 1
2017-06-30 1 1 1 1
2017-07-31 1 1 1 1
2017-08-31 1 1 1 1
2017-09-30 1 1 1 1
2017-10-31 1 1 1 1
2017-11-30 1 1 1 1
2017-12-31 1 1 1 1
2018-01-31 1 1 1 1
2018-02-28 1 1 1 1
2018-03-31 1 1 1 1
2018-04-30 1 1 1 1
2018-05-31 1 1 1 1
2018-06-30 1 1 1 1
2018-07-31 1 1 1 1
</code></pre>
<hr/>
<h2>Numpy切片</h2>
<p>创建一个自定义数组,定义零的放置位置</p>
<pre><code>i = np.array([
#A B C D
[1, 1, 0, 1], # Q1 -> Only column C is zero
[1, 0, 0, 0], # Q2 -> cols B, C, D are zero
[0, 0, 1, 1], # Q3 -> cols A, B are zero
[0, 1, 1, 0], # Q4 -> cols A, D are zero
])
q = df.index.quarter - 1
df * i[q]
A B C D
2016-10-31 0 1 1 0
2016-11-30 0 1 1 0
2016-12-31 0 1 1 0
2017-01-31 1 1 0 1
2017-02-28 1 1 0 1
2017-03-31 1 1 0 1
2017-04-30 1 0 0 0
2017-05-31 1 0 0 0
2017-06-30 1 0 0 0
2017-07-31 0 0 1 1
2017-08-31 0 0 1 1
2017-09-30 0 0 1 1
2017-10-31 0 1 1 0
2017-11-30 0 1 1 0
2017-12-31 0 1 1 0
2018-01-31 1 1 0 1
2018-02-28 1 1 0 1
2018-03-31 1 1 0 1
2018-04-30 1 0 0 0
2018-05-31 1 0 0 0
2018-06-30 1 0 0 0
2018-07-31 0 0 1 1
</code></pre>
<hr/>
<p>另一种观点认为,它是正确的季度工作。你知道吗</p>
<pre><code>i = np.array([
#A B C D
[1, 1, 0, 1], # Q1 -> Only column C is zero
[1, 0, 0, 0], # Q2 -> cols B, C, D are zero
[0, 0, 1, 1], # Q3 -> cols A, B are zero
[0, 1, 1, 0], # Q4 -> cols A, D are zero
])
q = df.index.quarter - 1
df.set_index(df.index.to_period('Q'), append=True).swaplevel(0, 1) * i[q]
A B C D
2016Q4 2016-10-31 0 1 1 0
2016-11-30 0 1 1 0
2016-12-31 0 1 1 0
2017Q1 2017-01-31 1 1 0 1
2017-02-28 1 1 0 1
2017-03-31 1 1 0 1
2017Q2 2017-04-30 1 0 0 0
2017-05-31 1 0 0 0
2017-06-30 1 0 0 0
2017Q3 2017-07-31 0 0 1 1
2017-08-31 0 0 1 1
2017-09-30 0 0 1 1
2017Q4 2017-10-31 0 1 1 0
2017-11-30 0 1 1 0
2017-12-31 0 1 1 0
2018Q1 2018-01-31 1 1 0 1
2018-02-28 1 1 0 1
2018-03-31 1 1 0 1
2018Q2 2018-04-30 1 0 0 0
2018-05-31 1 0 0 0
2018-06-30 1 0 0 0
2018Q3 2018-07-31 0 0 1 1
</code></pre>