Add weighting / other batteries included to newExampleDataset

[1beb]

> library(crunch); login()
> ds = newExampleDataset()
> weightVariables(ds) 
> NULL

When writing vignettes or creating examples for using R crunch, it would be useful if the example dataset came with a few more batteries included.

• A weighting variable
• Some variables that would traditionally be used as banners in a cross tabulation (age, gender, age cuts)

We, internally, already have a candidate dataset that is an extension of the current one.

[1beb]

If you could show me how to convert a ds to the json / csv formats you have in https://github.com/Crunch-io/rcrunch/tree/master/inst/example-datasets I would submit a PR for this

[malecki]

@1beb check out the new Schema endpoint — it provides a view of base data (not derivations) consistent with what is accepted in a create-dataset payload. Reminds me we should think about how to expose and interact with that endpoint here.

[1beb]

Ok, this looks about right so far:

toJSON(crGET("https://app.crunch.io/api/datasets/<DATASETID>/schema"))

But how do I generate the csv?

[malecki]

write.csv(as.data.frame(?

[gergness]

There's also exportDataset

[1beb]

The problem I'm having is that the weight variable doesn't appear to come through in the schema.

reprex:

library(crunch)
ds = newExampledataset() 
ds$weight = makeWeight(ds$q1 ~ c(0.75, 0.15, 0.10), name = "weight")
modifyWeightVariables(ds, "weight") # Assign "weight" as the weighting variable
exportDataset(ds, "example-dataset.csv") # has a column called weight as expected
# write.csv(as.data.frame(ds), "examples-dataset.csv") # has weight
toJSON(crGET("https://app.crunch.io/api/datasets/<DATASETID>/schema")) # no weight var

In addition, when trying to use the endpoint provided as the data it fails in createDatasetWithMediaAndData. I spent some time trying to manipulate the JSON result from the endpoint but it still looked different than the initial pets and it wasn't clear why it was being rejected.

[1beb]

Closing for now. Identified API issue where numeric variables output ids at schema end point making automatic round trips of data creation require a lot of munging.