0

I have an aggregated hierarchical dataset (aggregated per hour) as below

  TIME                level1 level2     n
  <dttm>              <fct>  <fct>  <int>
1 2019-11-29 00:00:00 A      a          1
2 2019-11-29 00:00:00 B      a          1
3 2019-11-29 00:00:00 B      b          2
4 2019-11-29 00:00:00 B      c          1
5 2019-11-29 01:00:00 A      a          2
6 2019-11-29 02:00:00 A      a          1
7 2019-11-29 02:00:00 B      a          1
8 2019-11-29 02:00:00 B      b          2
9 2019-11-29 02:00:00 B      c          1

I want to convert it to hts object and am following the instructions in hts manual and this post on cross validated: https://stats.stackexchange.com/questions/133894/get-groups-in-time-series-with-categorical-data-in-r-for-use-in-gts

I have convert the data into time series (mts) object as

df_mat <- ts(data.matrix(df_agg), frequency = 525960)

Frequency set to hour inline with the aggregation unit in the data frame.

As the number of distinct level1 and level2 can vary, i did not set nrow and ncol object.

After this, i am not very sure how to convert this object to hts. How to generate the arguments nodes, colnames and characters to be passed to the hts method ?

hts(y, nodes, bnames = colnames(y), characters)

Any suggestions ?

Thanks

Praveen
  • 2,137
  • 1
  • 18
  • 21
  • Can you describe the 'hts' object format that you need? – Fred Boehm Jan 19 '20 at 19:12
  • @FredBoehm, i want to be able to extract time series of level-1 (A,B ..etc) and level-2 (Aa, Ab, Ba, Bb, Bc) similar to the example in hts manual https://pkg.earo.me/hts/. Thanks – Praveen Jan 20 '20 at 02:58

0 Answers0