使用高频数据作为低频数据的代理

Question

I have two times series - annual and quarterly. Annual data ends in 2018, but quarterly data goes till 2019q3. What's the best way to combine the two, where Python checks what's the latest available quarterly and annual data and fills annual time series with the latest quarterly value?

This is what I have in mind:

Data_ann
2013 5.1
2014 3.2
2015 2.1
2016 2.2
2017 2.1
2018 4.2
2019 n/a

Data_qtr
...
2018q1 2.5 
2018q2 2.2
2018q3 3.7
2018q4 4.2
2019q1 1.2
2019q2 2.3
2019q3 n/a

and the result

2013 5.1
2014 3.2
2015 2.1
2016 2.2
2017 2.1
2018 4.2
2019 2.3

Answer 1

You can organize your data to use a DatetimeIndex . The yearly frame is then fine (if there's one row per year) but for the quarter DataFrame we need to take the last value in each year, accomplished with resample.last . combine_first gives us priority to the yearly DataFrame when we join them.

Data Prep

df_ann = pd.read_clipboard(header=None)
df_ann.columns = ['date', 'value']
df_ann['date'] = pd.to_datetime(df_ann['date'], format='%Y')
df_ann = df_ann.set_index('date')
#            value
#date             
#2013-01-01    5.1
#2014-01-01    3.2
#2015-01-01    2.1
#2016-01-01    2.2
#2017-01-01    2.1
#2018-01-01    4.2
#2019-01-01    NaN

df_qtr = pd.read_clipboard(header=None)
df_qtr.columns = ['date', 'value']
df_qtr['date'] = pd.to_datetime(df_qtr['date'])
df_qtr = df_qtr.set_index('date')
#            value
#date             
#2018-01-01    2.5
#2018-04-01    2.2
#2018-07-01    3.7
#2018-10-01    4.2
#2019-01-01    1.2
#2019-04-01    2.3
#2019-07-01    NaN

Code

df_ann.to_period('Y').combine_first(df_qtr.resample('Y').last().to_period('Y'))

      value
date       
2013    5.1
2014    3.2
2015    2.1
2016    2.2
2017    2.1
2018    4.2
2019    2.3