Add Synthetic CrossCat datasets #175
Conversation
| @@ -0,0 +1,4 @@ | |||
| col1 ~ bernoulli(id) | |||
There was a problem hiding this comment.
Doesn't the relation name here have to match the second column of the .obs file? Here it is "col1" but the .obs file uses "has_col1".
There was a problem hiding this comment.
Yeah this was a bad push. Fixed here and elsewhere.
| @@ -0,0 +1,4 @@ | |||
| col1 ~ stringcat[strings="a:b:c:d",delim=:](id) | |||
There was a problem hiding this comment.
In addition to the "has_" issue, there also appears to be an off by one error. Here, "col1" is the one that has a/b/c/d values, but it is "has_col0" in the .obs file that has those values.
Similarly for the other relations.
| NUM_SAMPLES_1 = 33 | ||
| NUM_SAMPLES_2 = 50 | ||
| NUM_SAMPLES = 100 | ||
|
|
There was a problem hiding this comment.
What do you think about adding some tests for this program?
There was a problem hiding this comment.
I'm adding integration tests in a later PR that will test the files (test some basic invariants, like creating two IRMs, etc)
Add four synthetic CrossCat datasets for each of the data types. Will add unit tests that verify that we get the appropriate clusterings with these later.