From 24c1a038397937883d0d30d53d50dd1cf86abdd5 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Francesc=20Mart=C3=AD=20Escofet?= Date: Mon, 24 Jun 2024 15:30:47 +0200 Subject: [PATCH] Add text about models --- docs/examples/example_basic.ipynb | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/docs/examples/example_basic.ipynb b/docs/examples/example_basic.ipynb index fe055eb..bf02295 100644 --- a/docs/examples/example_basic.ipynb +++ b/docs/examples/example_basic.ipynb @@ -215,6 +215,14 @@ "model a nuisance model, the initialization parameters for it are separated from the other\n", "nuisance parameters to allow a more understandable user interface, see the next code prompt.\n", "\n", + "In general, when initializing a MetaLearner the ``nuisance_model_factory`` parameter will\n", + "be used to create all the nuisance models which are not a propensity model, the\n", + "``propensity_model_factory`` will be used for the propensity model if the MetaLearner\n", + "contains one, and the ``treatment_model_factory`` will be used for the models predicting\n", + "the CATE. To see the models present in each MetaLearner type see\n", + "{meth}`~metalearners.metalearner.MetaLearner.nuisance_model_specifications` and\n", + "{meth}`~metalearners.metalearner.MetaLearner.treatment_model_specifications`.\n", + "\n", "In the {class}`~metalearners.RLearner` case, the ``nuisance_model_factory`` parameter will\n", "be used to create the outcome model, the ``propensity_model_factory`` will be used for the\n", "propensity model and the ``treatment_model_factory`` will be used for the model predicting\n",