Any example on how to use DSSFilterBuilder ?
![gnaldi62](https://us.v-cdn.net/6038231/uploads/lithium_attachments/n1462i304ED06A41271403.png)
Hi all,
we'd need to automatically change the dates related to the split between train and test of a visual ML model.
So far we've been able to get the raw details of the split via code similar to the following:
...
client = dataiku.api_client()
current_project = client.get_default_project()
my_model = current_project.get_ml_task('myAnalysisID', 'myTaskId')
split_params = my_model.get_settings().get_split_params().get_raw()
We then would need to change the values of a couple of dates in order to reflect a new split, but we are stuck here.
It seems we would need to use the class dataikuapi.dss.utils.DSSFilterBuilder, but we cannot find anywhere a reference on how to initialize nor to use it. Any ideas/suggestions ?
Thanks for your help. Best Regards.
Giuseppe
P.S.: we would like to keep the visual analysis rather than translating everything into a Python recipe.
Best Answer
-
Hi,
For updating an existing filter, it's easier, you can access the current definition of the filter in:
split_params.get_raw()["efsdTrain"]["filter"]["uiData"]
Ditto for "efsdTest".
"efsdTrain/efsdTest" are used for "Explicit Filter Single Dataset". If you are using "Explicit Filter Two Datasets", you need to use "eftdTrain" and "eftdTest"
For example, a "Pclass contains 1" filter translates to this value for uiData:
{ "mode": "&&", "conditions": [ { "input": "Pclass", "operator": "contains", "string": "1" } ] }
So switching that to "2" would be:
settings = mltask.get_settings() split_params = settings.get_split_params() split_params.get_raw()["efsdTrain"]["filter"]["uiData"]["conditions"][0]["string"] = "2" settings.save()
Hope this helps
Answers
-
gnaldi62 Partner, L2 Designer, Dataiku DSS Core Designer, Dataiku DSS & SQL, Dataiku DSS ML Practitioner, Dataiku DSS Core Concepts, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2022, Frontrunner 2022 Finalist, Frontrunner 2022 Winner, Frontrunner 2022 Participant, Neuron 2023 Posts: 79 Neuron
Hi Clémemt,
thanks for the quick reply. Best Regards.Giuseppe