Extract a column from crossfilter

Question

Heres my problem. I have a working dc.js based dashboard and some data within it. A column of the data contains text data (twitter info). Is it somehow possible to extract that specific column from crossfilter? My aim is to create some charts and the crossfilter containing the text data should feed into a d3 based word cloud so that i can do the drill down based filtering as well which dc and crossfilter provide out of the box. I tried a dimension.top(infinity) but that returns all the key value pairs in the data. I just need the values for a particular key across the whole data set. I hope my question makes some sense.

EDIT: More research reveals that the wordcloud will accept data in key value pair where the key is the word and value is its frequency of appearance. So i am guessing that will need to be implemented as well. If there is a ready to implement library out there kindly let me know as well. This changes things a bit as far as crossfilter is concerned.I need to throw this calculated key value pair (fit for the word cloud consumption) whenever a filter is triggered. How to go about it?

Looking forward to hearing from you all.

Best, Anmol

Ethan Jewett Ethan Jewett · Accepted Answer · 2016-01-09T18:10:57

Answer to the first part of the question: Probably dimension.top(Infinity) and then use an accessor to get the values you need. Not exactly efficient, but it is what it is.

Answer to the 2nd part of the question:

You need groupAll, I think. You want to take a tweet, generate an array of tokens (words), then generate a Crossfilter grouping that is a count per word, right? You can code your own custom crossfilter.dimension.groupAll reducers (if you want to do that, create a working example and I can probably cook it up). Or if you want to use Reductio:

tweetWords = data.dimension(function(d) { return d.tweetText.split(' '); });
wordCounts = tweetWords.groupAll();

reducer = reductio()
  .groupAll(function(d) {
    return d.tweetText.split(' ');
  })
  .count(true);

reducer(wordCounts);

wordCounts.all();

If you want to filter on this dimension you'll have to override the filter handler and check if the group key is in the dimension array for the record using a filterFunction.

Extract a column from crossfilter

1 Answers