processing data based on values in an object, which must be filtered based on key contents

gistfile1.js

// the issue here: We want to ignore census "label columns" for certain charts.
// We can identify these columns by the ".5" in their columnID. We're working
// from a `comparison.table.columns` object, which has columnIDs as keys. So
// the job here is to ignore all the columns in this object that have a "."
// in their key, and process the rest into chart data.


// one way is to grab a set of keys without ".", then make an object out of
// the columns that match the "good" keys, and process those for the chart

var filteredKeys = _.filter(_.keys(comparison.table.columns), function(k) { return k.indexOf('.') == -1 }),
    filteredColumns = _.pick(comparison.table.columns, filteredKeys)

_.each(filteredColumns, function(v, k) {
    // do the processing
});

// another way is to just send all the columns through the loop,
// and only process the ones that don't have a "." in their key.
// this feels simpler, so why does it feel more hacky?

_.each(comparison.table.columns, function(v, k) {
    if (k.indexOf('.') == -1) {
        // do the processing
    }
});

i have no problem with the latter (at line 22). If it feels hacky, I think it's because of the Bureau's casual approach to metadata as much as anything else. Maybe it also feels a little hacky because you have to depend on Underscore for the functional bits as opposed to having them as language built-ins (so you end up with more verbose code).

The slightly less hacky thing, if comparison.table or comparison.table.columns is a object beyond a mere array, would be to implement it as an object method, jus to encapsulate the behavior.