excluding "missing" values in reductio.avg()

Question

I was hoping to use reductio to compute averages within my crossfilter groups. My dataset includes missing values (represented by null) that I'd like to exclude when calculating the average. However, I don't see a way to tell reductio to exclude certain values, and it treats the null values as 0.

I wrote a custom reduce function to accomplish this without using reductio:

function reduceAvg(attr) {
    return {
        init: function() {
            return {
                count: 0,
                sum: 0,
                avg: 0
            };
        },
        add: function(reduction, record) {
            if (record[attr] !== null) {
                reduction.count += 1;
                reduction.sum += record[attr];
                if (reduction.count > 0) {
                    reduction.avg = reduction.sum / reduction.count;
                }
                else {
                    reduction.avg = 0;
                }
            }
            return reduction;
        },
        remove: function(reduction, record) {
            if (record[attr] !== null) {
                reduction.count -= 1;
                reduction.sum -= record[attr];
                if (reduction.count > 0) {
                    reduction.avg = reduction.sum / reduction.count;
                }
                else {
                    reduction.avg = 0;
                }
            }
            return reduction;
        }
    };
}

Is there a way to do this using reductio? Maybe using exception aggregation? I haven't fully wrapped my head around how exceptions work in reductio.

Answer 1

I think you should be able to average over 'myAttr' excluding null and undefined by doing:

reductio()
  .filter(function(d) { return d[myAttr] !== null && d[myAttr] !== undefined; })
  .avg(function(d) { return d[myAttr]; });

If that doesn't work as expected, please file an issue as it is a bug.