Timeline for Use the target variable during imputation?
Current License: CC BY-SA 4.0
10 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 3, 2023 at 11:53 | comment | added | dipetkov | You can't interpret feature importances causally. For one thing, the importances will be specific to the random forest model: it's the features that the RF uses to make its predictions. The true root cause of an event is not determined by what model you decide to use to analyze the data. | |
| Apr 2, 2023 at 18:17 | vote | accept | MarkH | ||
| Apr 2, 2023 at 10:10 | comment | added | MarkH | Feature importance for inferential interpretation (root cause analysis) | |
| Apr 2, 2023 at 10:01 | comment | added | Björn | Feature importance for what purpose? Inferential interpretation? Feature selection for prediction modeling? Something else? | |
| Apr 2, 2023 at 9:51 | comment | added | MarkH | I'm trying to build a model from which to derive feature importance. Predictions of this model play only a minor role. | |
| Apr 1, 2023 at 22:19 | comment | added | Björn | What are you trying to do? What you should do in missing data imputation actually depends on your setting. E.g. are you trying to answer causal questions? Are you trying to build a prediction model? | |
| Apr 1, 2023 at 20:02 | answer | added | EdM | timeline score: 2 | |
| Apr 1, 2023 at 15:43 | history | edited | MarkH | CC BY-SA 4.0 |
forgot one word "but"
|
| Apr 1, 2023 at 12:35 | history | edited | MarkH |
added missing-data tag
|
|
| Apr 1, 2023 at 11:51 | history | asked | MarkH | CC BY-SA 4.0 |