如何按季度分组并使用numpy从数组中计算平均值？

Question

我想利用numpy的从下方阵列计算每个季度的一组值的平均值：

Data = [{'date':'2015-01-01',value:5},{'date':'2015-02-01',value:6},{'date':'2015-03-01',value:7},{'date':'2015-04-01',value:8},{'date':'2015-05-01',value:9},{'date':'2015-06-01',value:10},{'date':'2015-07-01',value:11},{'date':'2015-08-01',value:12}] 

我想结果告诉我下面的：

• 对于Q1-15，平均为6
• 对于Q2-15，平均为9
• 对于Q3-15，平均为11.5

基于this stackoverflow question，我已经试过以下：

np = Data #I'm not sure how to read in data into numpy from an array in my structure 
np.resample('Q',how='mean') #I'm not sure if using 'Q' would group this by quarter 

Answer 1

我认为大熊猫能在这种情况下更好。我只会用你的简单例子来说明。

import pandas as pd # use recent version which has dt.quarter attr for time 
import json 
value = 'value' # to be able to read your Data string as json 
Data1 = json.dumps(Data) # need it to use read_json() method. 
a = pd.read_json(Data1) 
a[a['date'].dt.quarter == 1].mean() # 1st quarter 
a[a['date'].dt.quarter == 2].mean() # 2nd quarter 
a[a['date'].dt.quarter == 3].mean() # 3rd quarter