Sampling from a distribution

[verdverm]

I'd like to generate data where X% of users have a value set, perhaps having several (overlapping) options like linked accounts or something.

Is this possible or now a feature request?

[AlexMikhalev]

You can adjust the frequency of categorical fields:

{
  "type": "string",
  "categorical": {
    "USD": 8,
    "GBP": 1,
    "EUR": 1
  }
}

USD will appear eight times more likely than GBP/EUR.
You can sample from the distribution:

{
  "type": "number",
  "subtype": "f64",
  "range": {
    "low": 274.4,
    "high": 6597.5,
    "step": 0.01
  }
}

or sample from data source:

{
  "type": "datasource",
  "path": "json:descriptions.json",
  "cycle": true
}

where descriptions

[
  "Walmart Credit Purchase",
  "Amazon - Kindle reading tablet",
  "Apple store subscription",
  "Google store rental",
  "Apple accessories",
  "Amazon Kindle subscription"
]

see Tweaking individual fields
I am converting this issue to discussion for others' benefit.