Count records excluding duplicates [duplicate] - dc.js

Question

I am stuck at a unique problem involving dc.js and crossfilter. I have some data which i need to display using Number Charts powered by dc.js. However i found minimal documentation for the number charts and hence posting my query.
Here is the JSFiddle for what i have conceptualized so far.
I basically want to show the unique project count in box 1 which in this case would be 3, unique place count in box 2 which in this case would be 11 and the screen failure rate which would be 2/15*100 i.e. 15.3%
Currently i have made this working using jquery but thats just a hack. I would like to have these number charts based on cross table aggregation so that i can drill down on the data.
I have come across examples for reductions to calculate counts but they were for bar charts but in the number chart we need to have a value accessor for displaying data.
Can someone help me out please?
PS:
Here is the jquery code that i wrote. Dont know if this would be helpful.
$(document).ready(function() {
var baseURL = window.location.origin;
$.ajax({
url : baseURL + '/api/PlaceTable',
type : 'GET',
data : {},
async : true,
dataType : "json",
success : function(response) {
//Project Count
var projectIdCount = [];
for (i = 0; i < response.length; i++) {
if(response[i].Project != undefined){
if($.inArray(response[i].Project, projectIdCount) === -1){
projectIdCount.push(response[i].Project);
}
}
}
$('#number-box1').text(ProjectIdCount.length);
//Place Count
var placeIdCount = [];
for (i = 0; i < response.length; i++) {
if(response[i].Place != undefined){
if($.inArray(response[i].Place, placeIdCount) === -1){
placeIdCount.push(response[i].Place);
}
}
}
And for displaying a running sum of a column containing binary values i used this code, which worked in the number chart:
numberChart
.valueAccessor(function(x){ return +flag.groupAll().reduceCount().reduceSum(function(d) { return d.Flag; }).value();})
.group(ndx.groupAll());

Answer 1

The failure percentage calculation is a separate problem which I think you've asked elsewhere. To get the unique count, it is pretty easy to make a "fake groupAll" which returns the number of unique keys in its value method.
We'll also need to filter out the empty bins since crossfilter doesn't do that automatically.
function bin_counter(group) {
return {
value: function() {
return group.all().filter(function(kv) {
return kv.value > 0;
}).length;
}
};
}
var projectGroup = project.group();
projectCount
.valueAccessor(function(x){ return x;})
.group(bin_counter(projectGroup));
Updated fiddle here, still ignoring the failure% part:
http://jsfiddle.net/gordonwoodhull/vct0dzou/1/