From d7964c9cc38ca171c31b945ef5d075ca7889fd21 Mon Sep 17 00:00:00 2001
From: Peter Wu <peter@lekensteyn.nl>
Date: Wed, 21 May 2014 14:15:47 +0200
Subject: Option to ignore users with fewer than X tweets

---
 bubble.js     |  5 ++++-
 preprocess.js | 18 ++++++++++++++----
 2 files changed, 18 insertions(+), 5 deletions(-)

diff --git a/bubble.js b/bubble.js
index 773e481..81435e2 100644
--- a/bubble.js
+++ b/bubble.js
@@ -7,6 +7,8 @@
 var I_GOT_MONEY = 0;
 // whether to ignore lonely users
 var KILL_LONERS = true;
+// if positive, then users with fewer than this number of tweets will be ignored
+var USER_MIN_TWEET_COUNT = 0;
 // if true, enable sticky nodes unless Ctrl is held. If false, stick only if
 // ctrl is held (the inverse).
 var STICKY_DEFAULT = true;
@@ -138,7 +140,8 @@ function processData(data) {
     var infoPane = d3.select('#infobox');
 
     preprocess(data, {
-        kill_loners: KILL_LONERS
+        kill_loners: KILL_LONERS,
+        minTweetCount: USER_MIN_TWEET_COUNT
     });
     infoPane.select('.node-count').text(data.nodes.length);
     infoPane.select('.edge-count').text(data.edges.length);
diff --git a/preprocess.js b/preprocess.js
index 0bd13e5..7de2717 100644
--- a/preprocess.js
+++ b/preprocess.js
@@ -6,14 +6,14 @@
 'use strict';
 
 /* find user nodes and remove invalid edges */
-function filterEdges(data) {
+function filterEdges(data, ratelimit_max) {
     // map userID to nodes
     var users = {};
     data.nodes.forEach(function (user, i) {
         users[user.group] = user;
     });
 
-    var ratelimit_count = 0, ratelimit_max = 10;
+    var ratelimit_count = 0;
     function ratelimit() {
         return ++ratelimit_count <= ratelimit_max;
     }
@@ -37,7 +37,7 @@ function filterEdges(data) {
         }
         return !invalid;
     });
-    if (ratelimit_count > ratelimit_max) {
+    if (ratelimit_max > 0 && ratelimit_count > ratelimit_max) {
         console.log('Supressed', ratelimit_count, 'messages');
     }
 }
@@ -45,9 +45,19 @@ function filterEdges(data) {
 function preprocess(data, options) {
     console.log('Initial nodes count:', data.nodes.length);
     console.log('Initial edges count:', data.edges.length);
-    filterEdges(data);
+    filterEdges(data, 10);
     console.log('Valid edges count:', data.edges.length);
 
+    if (options.minTweetCount > 0) {
+        /* filter away users with almost no tweets */
+        data.nodes = data.nodes.filter(function (node) {
+            return node.tweetCount >= options.minTweetCount;
+        });
+        console.log('Nodes count (ignoring users with fewer than',
+            options.minTweetCount, 'tweets):', data.nodes.length);
+        filterEdges(data, 0);
+    }
+
     // find all related users by userID
     var hasRelations = {};
     data.edges.forEach(function (link) {
-- 
cgit v1.2.1