Add exploratory data analysis to come up with a real-data example.

[ml_measurement_error_public.git] / civil_comments / perspective_json_to_csv.sh

diff --git a/civil_comments/perspective_json_to_csv.sh b/civil_comments/perspective_json_to_csv.sh
new file mode 100755 (executable)
index 0000000..8e2aa13
--- /dev/null
+++ b/civil_comments/perspective_json_to_csv.sh
@@ -0,0 +1,4 @@
+#!/usr/bin/bash
+header=id,identity_attack_prob,insult_prob,profanity_prob,severe_toxicity_prob,threat_prob,toxicity_prob
+echo "$header" > $2
+cat $1 | jq -r '[.id, .attributeScores.IDENTITY_ATTACK.summaryScore.value, .attributeScores.INSULT.summaryScore.value, .attributeScores.PROFANITY.summaryScore.value,.attributeScores.SEVERE_TOXICITY.summaryScore.value, .attributeScores.THREAT.summaryScore.value, .attributeScores.TOXICITY.summaryScore.value] | @csv' >>  $2