Filter out noise

whotracksme/data/loader.py

@@ -352,6 +352,7 @@ def get_presence_by_site_category(self, id):
             WHERE st.month = ?
                 AND st.country = ?
                 AND tracker = ?
+                AND site_proportion > 0.05
             GROUP BY sites_data.category
             ORDER BY count DESC
         ''', (self.last_month, self.region, id))
@@ -468,6 +469,7 @@ def get_tracker_list(self, site, month=None):
             LEFT JOIN companies ON companies.id = trackers.company_id
             WHERE month = ? AND sites_trackers_data.country = ?
                 AND site = ? AND category != "extensions"
+                AND site_proportion > 0.05
             ORDER BY site_proportion DESC
         '''
         return map(SiteTrackerEntry._make, self.db.connection.execute(query, (month or self.last_month, self.region, site)))
@@ -514,6 +516,7 @@ def get_site_tracker_categories(self, site, month=None):
             JOIN categories ON trackers.category_id = categories.id
             WHERE month = ? AND sites_trackers_data.country = ?
                 AND site = ?
+                AND site_proportion > 0.05
             GROUP BY category
             ORDER BY frequency DESC
         '''

whotracksme/website/build/trackers.py

@@ -42,6 +42,7 @@ def tag_cloud_data(tracker_id, data):
     WHERE sites_trackers_data.month = ?
         AND sites_trackers_data.country = ?
         AND sites_trackers_data.tracker = ?
+        AND site_proportion > 0.05
     ORDER BY tracker_proportion DESC
     ''', (data.trackers.last_month, data.trackers.region, tracker_id))