file naming

[aremazeilles]

In the document, we repeat always something like:

Name of the file: subject_N_run_R_gaitEvents where N = subject’s number and R= run number. Use appropriate leading zeros for R and N to ensure proper ordering of files.

In #4 discussion, we also mentioned that conditions could also change during the experimentation. So I think we should provide at the bottom a definition common to all:

• We assume an experiment can involve N subjects
• Several conditions C can be tested in a given experiment (like slope of 10 deg, slope of 15, 20, ...)
• The subject conduct M repetitions or runs of the same conditions

So that the captured files should follow the pattern: subject_N_cond_C_run_R_dataname.csv, together with the testbed configuration files cond_C_testbed.yaml.

If one of the factor elements is not multiple in the protocol / Experiment, then it can be omitted from the naming:

• if only a subject is involved, then subject_N_ files can be omitted and the datafiles are deduced to: cond_C_run_R_dataName.csv
• if a single configuration is tested, then a single testbed.yaml file is provided, and data files are named: subject_N_run_R_dataName.csv
• if a single run s captured per condition (even though this is not the preferred scheme), then datafiles are named: cond_C_subject_N_dataName.csv
• In a particuliar experimen involving a unique subject, with a single run, and unique configuration, then the datafilenames would be reduced to dataName.csv

Open points still to clarify?

• what should be the outter field: condition or subject (in the naming for instance)
• Should the configuration parameters common to all subject (all go through cond1, then cond2, ...), or can we have some parameters that change from a subject to another?
• is the name cond appropriate ? Or configuration ? .... I used the term controlled_variables in the template spec, but I don't have clear what would be the most evident here.

ping @alfonsotecnalia

[alfonsotecnalia]

* is the name `cond` appropriate ? Or `configuration` ?

I was already writing the code for the PIManager for handling this. I have had the same problem with the nomenclature. The name of the branch is experiment_condition but inside the code I look for files matching config_XX. I would discard controlled_variables because the problems of understading we are having. If we are thinking in a unique file for both conditions and equipments configurations I would say to use config or configuration , if different then cond or condition.

* what should be the outter field: condition or subject (in the naming for instance)

I would say it is more important the condition than the subject so condition should go in the outer field. At some momment I was thiking in includig the subject in the file and avoid having 'subject_S' in all the files but that could lead to a big number of conditions files.

* Should the configuration parameters common to all subject (all go through cond1, then cond2, ...), or can we have some parameters that change from a subject to another?

I would say one common for all the subjects. However not all the subjects would go through all the conditions. In some experiments we could have 6 different conditions (6 slopes) and have data for less conditions (because the user is not able to climb stepper).

[aremazeilles]

So we should go for cond, is this ok?

[alfonsotecnalia]

yes

[aremazeilles]

OK, working on it