diff --git a/valor-sandbox/optimized_implementation.py b/valor-sandbox/optimized_implementation.py
index 81c485cdf..146a5b12e 100644
--- a/valor-sandbox/optimized_implementation.py
+++ b/valor-sandbox/optimized_implementation.py
@@ -118,9 +118,7 @@ def _calculate_pr_curves_optimized(
     )
 
     pr_curve_counts_df.fillna(0, inplace=True)
-
-    '''
-    Pretty sure there's a bug with accuracy, it assumes that each `datum_id` has every `label_key`.
+    pr_curve_counts_df["total_datums"] = pr_curve_counts_df["label_key"].map(total_datums_per_label_key.to_dict())
 
     pr_curve_counts_df["precision"] = pr_curve_counts_df["true_positives"] / (
         pr_curve_counts_df["true_positives"]
@@ -133,13 +131,12 @@ def _calculate_pr_curves_optimized(
     pr_curve_counts_df["accuracy"] = (
         pr_curve_counts_df["true_positives"]
         + pr_curve_counts_df["true_negatives"]
-    ) / len(unique_datum_ids)
+    ) / pr_curve_counts_df["total_datums"]
     pr_curve_counts_df["f1_score"] = (
         2 * pr_curve_counts_df["precision"] * pr_curve_counts_df["recall"]
     ) / (pr_curve_counts_df["precision"] + pr_curve_counts_df["recall"])
 
     # any NaNs that are left are from division by zero errors
     pr_curve_counts_df.fillna(-1, inplace=True)
-    '''
 
     return pr_curve_counts_df
\ No newline at end of file