read.ia: read instar analytics export
In rluech/tv-clone: TV package

Define the Report and its Export in Instar Analytics according to these rules to allow a predictable file format ready for automatic processing

1	read.ia(file)

Always display the report with only one header row. This is the 'long' format, any further variable is added as rows not as columns, e.g.:

Timebands | Dates | Sample | Universe | AvRch(000) | ... 'Whole Day' | '01.01.2017' | 1995 | 3351.017 | 2784.1940 | ... 'Whole Day' | '02.01.2017' | 1995 | 3351.017 | 2784.1940 | ... 'Whole Day' | '03.01.2017' | 1995 | 3351.017 | 2784.1940 | ...

This has to be defined in the 'period' sheet in Instar. If the job ist saved, this display formating will be retained the next time the job is executed. The export table will be written the same way as the period table is formatted.

Make sure to not display any Totals in the report. You can disable Totals in the 'Template' sheet.

If there are unwanted attributes in the period table select from the menue report > attributes > channel to deselect columns. Only export columns that you need

Which seperator to use should not matter when using data.table::fread(), however ; or , is recommended

'Print labels' is needed otherwise header as well as grouping variable will be missing completely. Use 'repeate every row/column'. Otherwise you have to fill empty cells in R using local observation carried forward na.locf().

numbers should be exported with the highest possible precision which is 9 decimals. This can be set in options.

Exporting to .txt is probably most predictable. Only Export the report, do not export Technical Card, which would append in the same file.

For reproducebility save the job and batch, then Technical Card is not needed

A data.table. column header should match the variable names used in the TV package. Variable values should also match thode used in the TV package, e.g in id$lab[variable].

rluech/tv-clone documentation built on Jan. 7, 2022, 12:27 a.m.