Skip to content

Commit

Permalink
add a practical implication to the conclusion
Browse files Browse the repository at this point in the history
  • Loading branch information
rfl-urbaniak committed Dec 19, 2024
1 parent 349a801 commit a264ba6
Showing 1 changed file with 3 additions and 1 deletion.
4 changes: 3 additions & 1 deletion docs/source/explainable_sir.ipynb
Original file line number Diff line number Diff line change
Expand Up @@ -1744,7 +1744,9 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"`SearchForExplanation` can be used in any `ChiRho` causal model to investigate the causal role of variables of interest in outome variables having the values that they do in particular runs. This is achieved by transforming the causal model into one that allows us to investigate multiple causal settings with \"necessity\" and \"sufficiency\" interventions and their impact on the outcome variable across different possible worlds. Looking back at the heatmap, \"Counterfactual heatmap of overshoot under necessity and sufficiency interventions\" with the accompanying explanation illustrates how paying attention to both conjoining sufficiency and necessity and to searching through actual contexts to be fixed allows us to break the *prima facie* symmetries which makes the usual but-for analysis blind to such a fine-grained analysis. Moreover, the obtained samples can be used to answer more specific causal queries if need be."
"`SearchForExplanation` can be used in any `ChiRho` causal model to investigate the causal role of variables of interest in outome variables having the values that they do in particular runs. This is achieved by transforming the causal model into one that allows us to investigate multiple causal settings with \"necessity\" and \"sufficiency\" interventions and their impact on the outcome variable across different possible worlds. Looking back at the heatmap, \"Counterfactual heatmap of overshoot under necessity and sufficiency interventions\" with the accompanying explanation illustrates how paying attention to both conjoining sufficiency and necessity and to searching through actual contexts to be fixed allows us to break the *prima facie* symmetries which makes the usual but-for analysis blind to such a fine-grained analysis. Moreover, the obtained samples can be used to answer more specific causal queries if need be.\n",
"\n",
"A practical policy-making implication is that developing causal models for problems at hand and employing a strategy analogous to the one discussed in this notebook is a promising strategy for a fine-grained and probabilistically aware responsibility attribution."
]
}
],
Expand Down

0 comments on commit a264ba6

Please sign in to comment.