fix: use top 50, top 100 is useless

patrickhulce · Mar 14, 2019 · 37b69df · 37b69df
1 parent 647a174
commit 37b69df
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 20 deletions.
diff --git a/bin/find-entities-to-label.js b/bin/find-entities-to-label.js
@@ -86,14 +86,14 @@ function computeAllStats(dataset) {
     .reverse()
     .value()
 
-  const top100Occurrences = _.sumBy(
-    sortedEntityData.filter(e => e !== homelessMegaEntity).slice(0, 100),
+  const top50Occurrences = _.sumBy(
+    sortedEntityData.filter(e => e !== homelessMegaEntity).slice(0, 50),
     'totalOccurrences',
   )
 
   return {
     sortedEntityData,
-    top100Occurrences,
+    top50Occurrences,
     homelessGrouped,
     totalEntityOccurrences,
   }
@@ -118,7 +118,7 @@ function computeChangesSinceLast(currentDataset, lastDataset) {
 const currentDatasetStats = computeAllStats(CURRENT_DATASET)
 const lastDatasetStats = computeAllStats(LAST_DATASET)
 
-const {top100Occurrences, homelessGrouped, totalEntityOccurrences} = currentDatasetStats
+const {top50Occurrences, homelessGrouped, totalEntityOccurrences} = currentDatasetStats
 
 const changesSinceLast = computeChangesSinceLast(currentDatasetStats, lastDatasetStats)
 
@@ -160,20 +160,19 @@ console.log(
 )
 
 console.log(
-  'Top 100 Entities representing',
-  ((top100Occurrences / GLOBAL_OCCURRENCES) * 100).toFixed(2),
+  '3rd parties representing',
+  ((THIRD_PARTY_REQUESTS / GLOBAL_OCCURRENCES) * 100).toFixed(2),
   '% of total requests',
 )
-
 console.log(
   'Entities representing',
   ((totalEntityOccurrences / THIRD_PARTY_REQUESTS) * 100).toFixed(2),
   '% of 3rd parties',
 )
 console.log(
-  '3rd parties representing',
-  ((THIRD_PARTY_REQUESTS / GLOBAL_OCCURRENCES) * 100).toFixed(2),
-  '% of total requests',
+  'Top 50 Entities representing',
+  ((top50Occurrences / totalEntityOccurrences) * 100).toFixed(2),
+  '% of all entity requests',
 )
 
 console.log('Finished processing', datasetFiles[0])
diff --git a/bin/merge-origins-with-entities.js b/bin/merge-origins-with-entities.js
@@ -93,12 +93,12 @@ function computeAllStats(dataset) {
     .reverse()
     .value()
 
-  const top100ExecutionTime = _.sumBy(
-    sortedEntityData.filter(e => e !== homelessMegaEntity).slice(0, 100),
+  const top50ExecutionTime = _.sumBy(
+    sortedEntityData.filter(e => e !== homelessMegaEntity).slice(0, 50),
     'totalExecutionTime',
   )
 
-  return {sortedEntityData, top100ExecutionTime, homelessGrouped, totalEntityExecutionTime}
+  return {sortedEntityData, top50ExecutionTime, homelessGrouped, totalEntityExecutionTime}
 }
 
 function computeChangesSinceLast(currentDataset, lastDataset) {
@@ -122,7 +122,7 @@ const lastDatasetStats = computeAllStats(LAST_DATASET)
 
 const {
   sortedEntityData,
-  top100ExecutionTime,
+  top50ExecutionTime,
   homelessGrouped,
   totalEntityExecutionTime,
 } = currentDatasetStats
@@ -180,20 +180,19 @@ console.log(
 )
 
 console.log(
-  'Top 100 Entities representing',
-  ((top100ExecutionTime / GLOBAL_EXECUTION_TIME) * 100).toFixed(2),
+  '3rd parties representing',
+  ((THIRD_PARTY_EXECUTION_TIME / GLOBAL_EXECUTION_TIME) * 100).toFixed(2),
   '% of total script execution',
 )
-
 console.log(
   'Entities representing',
   ((totalEntityExecutionTime / THIRD_PARTY_EXECUTION_TIME) * 100).toFixed(2),
   '% of 3rd parties',
 )
 console.log(
-  '3rd parties representing',
-  ((THIRD_PARTY_EXECUTION_TIME / GLOBAL_EXECUTION_TIME) * 100).toFixed(2),
-  '% of total script execution',
+  'Top 50 Entities representing',
+  ((top50ExecutionTime / totalEntityExecutionTime) * 100).toFixed(2),
+  '% of total entity execution',
 )
 
 fs.writeFileSync(