Fix double counting in two sided bootstrap confidence intervals (#1126)

* fixed double counting in bootstrap conf interval

Signed-off-by: Amit Sharma <[email protected]>

* fixed double counting in bootstrap conf interval

Signed-off-by: Amit Sharma <[email protected]>

* formatted correctly

Signed-off-by: Amit Sharma <[email protected]>

---------

Signed-off-by: Amit Sharma <[email protected]>

Author: amit-sharma

dowhy/causal_estimator.py

@@ -350,9 +350,11 @@ def _estimate_confidence_intervals_with_bootstrap(
         bootstrap_variations = [bootstrap_estimate - estimate_value for bootstrap_estimate in bootstrap_estimates]
         sorted_bootstrap_variations = np.sort(bootstrap_variations)
 
-        # Now we take the (1- p)th and the (p)th variations, where p is the chosen confidence level
-        upper_bound_index = int((1 - confidence_level) * len(sorted_bootstrap_variations))
-        lower_bound_index = int(confidence_level * len(sorted_bootstrap_variations))
+        # Now we take the (1-p)/2 th and the 1-(1-p)/2 th variations, where p is the chosen confidence level
+        left_fraction = (1 - confidence_level) / 2
+        right_fraction = 1 - left_fraction
+        upper_bound_index = int(left_fraction * len(sorted_bootstrap_variations))
+        lower_bound_index = int(right_fraction * len(sorted_bootstrap_variations))
 
         # Get the lower and upper bounds by subtracting the variations from the estimate
         lower_bound = estimate_value - sorted_bootstrap_variations[lower_bound_index]