5. 堆叠面积图，以发现趋势

  1. #  读取meetup_groups数据集
  2.  In[66]: meetup = pd.read_csv('data/meetup_groups.csv', 
  3.                               parse_dates=['join_date'], 
  4.                               index_col='join_date')
  5.          meetup.head()
  6. Out[66]:

5. 堆叠面积图，以发现趋势 - 图1

  1. #  算出每周加入每个组的人
  2.  In[67]: group_count = meetup.groupby([pd.Grouper(freq='W'), 'group']).size()
  3.          group_count.head()
  4. Out[67]: join_date   group   
  5.          2010-11-07  houstonr     5
  6.          2010-11-14  houstonr    11
  7.          2010-11-21  houstonr     2
  8.          2010-12-05  houstonr     1
  9.          2011-01-16  houstonr     2
  10.          dtype: int64
  1. #  将数据表unstack
  2.  In[68]: gc2 = group_count.unstack('group', fill_value=0)
  3.          gc2.tail()
  4. Out[68]:

5. 堆叠面积图，以发现趋势 - 图2

  1. #  做累积求和
  2.  In[69]: group_total = gc2.cumsum()
  3.          group_total.tail()
  4. Out[69]:

5. 堆叠面积图，以发现趋势 - 图3

  1. #  将每行分开，已找到其在总数中的百分比
  2.  In[70]: row_total = group_total.sum(axis='columns')
  3.          group_cum_pct = group_total.div(row_total, axis='index')
  4.          group_cum_pct.tail()
  5. Out[70]:

5. 堆叠面积图，以发现趋势 - 图4

  1. #  话堆叠面积图
  2.  In[71]: ax = group_cum_pct.plot(kind='area', figsize=(18,4),
  3.                                  cmap='Greys', xlim=('2013-6', None), 
  4.                                  ylim=(0, 1), legend=False)
  5.          ax.figure.suptitle('Houston Meetup Groups', size=25)
  6.          ax.set_xlabel('')
  7.          ax.yaxis.tick_right()
  9.          plot_kwargs = dict(xycoords='axes fraction', size=15)
  10.          ax.annotate(xy=(.1, .7), s='R Users', color='w', **plot_kwargs)
  11.          ax.annotate(xy=(.25, .16), s='Data Visualization', color='k', **plot_kwargs)
  12.          ax.annotate(xy=(.5, .55), s='Energy Data Science', color='k', **plot_kwargs)
  13.          ax.annotate(xy=(.83, .07), s='Data Science', color='k', **plot_kwargs)
  14.          ax.annotate(xy=(.86, .78), s='Machine Learning', color='w', **plot_kwargs)
  15. Out[71]: Text(0.86,0.78,'Machine Learning')

5. 堆叠面积图，以发现趋势 - 图5

更多

  1. #  用饼图查看每组随时间的分布情况
  2.  In[72]: pie_data = group_cum_pct.asfreq('3MS', method='bfill') \
  3.                                  .tail(6).to_period('M').T
  4.          pie_data
  5. Out[72]:

5. 堆叠面积图，以发现趋势 - 图6

  1.  In[73]: from matplotlib.cm import Greys
  2.          greys = Greys(np.arange(50,250,40))
  4.          ax_array = pie_data.plot(kind='pie', subplots=True, 
  5.                                   layout=(2,3), labels=None,
  6.                                   autopct='%1.0f%%', pctdistance=1.22,
  7.                                   colors=greys)
  8.          ax1 = ax_array[0, 0]
  9.          ax1.figure.legend(ax1.patches, pie_data.index, ncol=3)
  10.          for ax in ax_array.flatten():
  11.              ax.xaxis.label.set_visible(True)
  12.              ax.set_xlabel(ax.get_ylabel())
  13.              ax.set_ylabel('')
  14.          ax1.figure.subplots_adjust(hspace=.3)
  15. Out[73]:

5. 堆叠面积图，以发现趋势 - 图7