causalBoosting returns matrix for prediction

[MaximilianFranz]

Hey Scott,

thanks for the awesome package! I've noticed that when I call predict(causalBoostingModel, newx), it returns a matrix of shape (size_newx, num_trees) with num_trees being the number of trees specified when fitting causal boost.
How should I deal with these return values, as I only want the treatment effect for each instance in newx? Can I just take the mean?

Thanks for your answer already!
Best,
Max

[saberpowers]

Max,

I'm sorry for not replying to your message back in August. It was a busy time at my full-time job, and then I forgot to reply. I assume you're no longer invested in an answer to this question, but here it is any way.

Interestingly, I do not get a matrix of shape (size_newx, num_trees) when I call predict(causalBoostingModel, newx). Rather, I get a vector of length size_newx. See, for example, the example provided in the documentation for causalBoosting():

n = 100 # number of training-set patients to simulate
p = 10  # number of features for each training-set patient
x = matrix(rnorm(n * p), nrow = n, ncol = p) # simulate covariate matrix
tx_effect = x[, 1] + (x[, 2] > 0) # simple heterogeneous treatment effect
tx = rbinom(n, size = 1, p = 0.5) # random treatment assignment
y = rowMeans(x) + tx * tx_effect + rnorm(n, sd = 0.001) # simulate response
fit_cb = causalBoosting(x, tx, y, num.trees = 500)
pred_cb = predict(fit_cb, newx = x, num.trees = 500)

Let me know if you're able to provide a minimal example that produces the behavior you're describing, and I'll take a look.

Scott