DayZSynthesizer get_parameters()

Synthetic Data Creation
1

Hi,
I am still looking around the multi table DayZ synthesizer and I was wondering why you guys have the get_parameters function and not a set_parameters function where we can load like a JSON file explaining all the parameters.
In some use cases we want to work with a lot of tables (let’s say 50 tables) and I have to add a lot of parameters. Such as categorical values, and numerical bounds. I feel like it is easier to create a dictionnary or a JSON file with all the params that looks like the one when we use the get_parameters() function than having to write/execute all the .set_category_values and .set_numerical_bounds.
I feel like it would be a nice feature to include in case we can’t to save time or to automate the whole process better.
At the moment I work on 5 - 10 different tables that has categorical values. The only way to ‘automate’ the process is to create a ‘for’ loop to set the parameters. It works fine but I feel like I can be better.
Bests,
Charles

2

Hi Charles, I can definitely empathize with the need to streamline how parameters are defined and passed to SDV! To help us better understand your use case, I have a few questions if you don’t mind:

  • Do you have some existing, representative data that you might be able to use for SDV to use in model training? I noticed that you’re using DayZSynthesizer so I thought I’d ask to gain more context into your data availability.
  • What’s your iteration workflow like for ‘discovering’ & defining the metadata and parameters you want to use here?

Either way, I’m tagging this as a feature request as we learn more from you!

3

Hi Srini,
I actually don’t have any data other than some tests metadata that I made. I am working on an app that creates tables and I wanted to create data in order to test the new import features of the app (I created a little converter that turns my app’s metadata into SDV format metadata). So the only thing that I have are the metadata.
For example I created a patient-observations scenario where we have 2 tables connected. In the patient table I have :

  • id : sdtype = id/primary_key
  • first name : sdtype = first_name
  • last name : sdtype = last_name
  • address : sdtype = street_address
  • age : sdtype = numerical/float
  • gender: sdtype = categorical
  • phone number : sdtype = phone_number
  • date of birth : sdtype = datetime/%YY%mm%DD
    And in the observation let’s say I have
  • id : sdtype = id/primary_key
  • patient.id : sdtype = id/foreign_key
  • height : sdtype = numerical/float
  • weight : sdtype = numerical/float
  • observations : sdtype = text

And in this case it’s really easy to set the DayZ parameters.
Like for age like set_numerical_bounds : between 10 - 110

  • height : 110cm - 230cm
  • weight : 30kg-200kg
    and also set category values for gender like male, female or other.

What if I had more conditions, more parameters like physiological records and other stuff ? It all comes down to realism. I want to add a lot of constraint to have realistic data maybe for later use such as demos or other stuff. But for now I’m just limit testing the whole package. And since I created some sort of converter I wanted to see if I could do the same thing for all the parameters.

4

Got it, this makes sense! Thanks for the context!