-
Suppose both X and W are continuous. Suppose X, W have very high correlation, with 0.95 correlation coefficient. The tree is built on X, where within a leaf the nodes have very similar X. Because of t…
-
On my current dataset, OOB predictions for causal forest are all NaN. y.hat is also all NaN. There are around 5000 observations split 50/50 between treatment and control. All features are integers, al…
-
Following code on the homepage for "Add confidence intervals for heterogeneous treatment effects; growing more trees is now recommended", I am trying to get tau(X1) while holding other X at 0. I stand…
-
Hello,
I am working through the ORF paper (arXiv:1806.03467v2 [cs.LG] 12 Jul 2018) and I have a small question that I couldn't figure out on my own.
What are the results from the simulation whe…
-
I am trying specifications with and without specifying clusters for the causal_tree function, with the same dataset.
It seems when clusters are specified, the computations time (for causal_tree and…
-
**Description of the bug**
The function `test_calibration()` fails when either `W` or `Y` are a named matrix column. The problem is in the code:
```r
DF = data.frame(target = forest$Y.orig - forest…
-
Currently, we don't validate the user-provided observation objects in R like `Y` and `W`. We should perform at least the following validation:
- Make sure that they are numeric vectors. Currently the…
-
**Description of the bug**
A) Tuning via tune_causal_forest(...) produce different results compared to tuning via causal_forest(...,tune.parameters = T). These tunning operations produce different …
-
**Description of the bug**
I have a dataset that is 2,000,000X34 with a binary outcome, binary treatment, and clustered data. The causal_tree finishes the estimation with 1000 trees without any erro…
-
This is not a short term bug, but it will become an issue when macOS Catalina is released this fall. macOS is now split across two volumes (system and data) with a read-only system volume. This means …