EconML

EconML copied to clipboard

MemoryError in econml.dml

2

I have a dataset with about 6.3 million observations (rows), 10 treatments and 4 effect modifiers. I am using econml.dml.LinearDML module like this: ` est = LinearDMLCateEstimator(model_t =MultiOutputRegressor(CatBoostRegressor()), model_y =...

hhu1

Question about a nearly perfect propensity model in doubly robust learning

6

As stated in the title, if the propensity model in doubly robust learning is nearly perfect, then the inverse propensity weight would be extremely high, causing Y^{DR}_{i,t} to be inaccurate...

DonCorle0ne

This is not intended to be merged, but to allow us to track down the cause of the crashes in test_tree

kbattocchi

Not intended to merge, just to look at memory usage.

kbattocchi

Test install

2

This PR is just to test the interactions of pip and numpy 1.20, and whether those break either us or our dependencies. It is not intended ever to be merged.

kbattocchi

Relax tensorflow version limit

7

kbattocchi

Native build proof of concept

2

This PR is not intendended to be merged, but to figure out how to get our build process working with the new native changes

kbattocchi

Sphinx 3.2 checks default argument specifications more stringently (see e.g. https://github.com/sphinx-doc/sphinx/issues/8088), so the build was failing with that version of Sphinx. These changes should address the issue, except that our...

kbattocchi

Basic monte carlo framework for simulating dgps and methods

1

Built on top of https://github.com/vsyrgkanis/plugin_regularized_estimation Changes made to to fit econml's structure Documentation upcoming

v-asatha

kbattocchi