@@ -6,6 +6,37 @@ DoWhy is hosted on GitHub.
6
6
You can browse the code in a html-friendly format `here
7
7
<https://github.com/Microsoft/dowhy> `_.
8
8
9
+
10
+ v0.7: Better Refuters for unobserved confounders and placebo treatment
11
+ ----------------------------------------------------------------------
12
+ * **[Major] ** Faster backdoor identification with support for minimal adjustment, maximal adjustment
13
+ or exhaustive search. More test coverage for identification.
14
+
15
+ * **[Major] ** Added new functionality of causal discovery [Experimental].
16
+ DoWhy now supports discovery algorithms from external libraries like CDT.
17
+ `[Example notebook] <https://github.com/microsoft/dowhy/blob/master/docs/source/example_notebooks/dowhy_causal_discovery_example.ipynb >`_
18
+
19
+ * **[Major] ** Implemented ID algorithm for causal identification. [Experimental]
20
+
21
+ * Added friendly text-based interpretation for DoWhy's effect estimate.
22
+
23
+ * Added a new estimation method, distance matching that relies on a distance
24
+ metrics between inputs.
25
+
26
+ * Heuristics to infer default parameters for refuters.
27
+
28
+ * Inferring default strata automatically for propensity score stratification.
29
+
30
+ * Added support for custom propensity models in propensity-based estimation
31
+ methods.
32
+
33
+ * Bug fixes for confidence intervals for linear regression. Better version of
34
+ bootstrap method.
35
+
36
+ * Allow effect estimation without need to refit the model for econml estimators
37
+
38
+ Big thanks to @AndrewC19, @ha2trinh, @siddhanthaldar, and @vojavocni
39
+
9
40
v0.6: Better Refuters for unobserved confounders and placebo treatment
10
41
----------------------------------------------------------------------
11
42
0 commit comments