Add Timeseries to datapackage

[FelixMau]

For Timesensitive facades a column is required in tabular that is reffering to a column in another Dataset that contains the necessary Timeseries for that facade. This column is usually called profile. Example:

Projekt-Verzeichnis
├── datapackage.json
├── scalars
│ ├── wind.csv
│ └── ...
└── sequences
├── wind_profile.csv
└── ...

wind.csv:	name	type	carrier	tech	capacity	bus	marginal_cost	profile
wind-onshore	volatile	wind	onshore	50	electricity	0	onshore-profile

wind_profile.csv:

timeindex	onshore-profile
2011-01-01T00:00:00Z	0.147532
2011-01-01T01:00:00Z	0.184181
2011-01-01T02:00:00Z	0.223937

In order to achieve the desired result, the following steps should be implemented:

• [x] Add profile column to facade Adapter (profile might have a different name in some facades).
• [x] Refactor timeseries Dataframe to have timeindex as index and values within the same column as it is in the profile column. Use timeindex_start timeindex_end and timeindex_resolution to create the correct timeindex
• [x] Matching of scalar row and time series profile (by year and region).
• [x] times series are to be written into csv. Multiple csv per facade as there are multiple periods/years. Timeseries could also be appended and put into one csv. (This might be better)
• [ ] Write the name of the file to foreign keys (this may happen also via an helper dictionary that could be saved somewhere to call later or added as parameter to the dataadapter class for a call of infer_metadata

[nailend]

@FelixMau I didn't get that far, as I was working with the wrong example.

I added a first draft for the time series refactoring. This still needs to be put into a separate function. Matching of scalar rows and time series does not happen yet. But they can be matched via region and year parameter. Therefore, these parameters can be used as identifier in the profile column. Timer series for each period/year are saved separately into the dictionary parametrized_sequences for now. But I guess it's better to append them in one data frame.

[FelixMau]

Thank you very much. I'll just put your input into a little checklist that will help me selforganising:

• [x] time series refactoring in seperate function
• [x] Match scalars to time series
• [x] Merge multiperiods into one Dataframe

[FelixMau]

To map the time series to scalars, I suggest the following pattern:

• Refactor the time series using the refactor_timeseries function.
• The facade adapters that include a time series will have an additional variable called profiles, which is a list of column names for profiles.
• If there is more than one time series or entries in profiles, a warning will be raised, and no time series will be processed (this feature is not yet implemented).
• When creating the scalar data frames, add another column whose name comes from the adapter's profiles variable.
• The entries of this profile column will be created using original name of the time series column + _ + region from the scalar.

To use this approach, it is important to name time series columns using descriptive and singular names.