Nest data containing arrays with d3.nest

Question

I am using d3 to make some pretty graphs, all of the same raw data. This raw data, however, contains arrays.

var data = [
{name: "Alfred", age: "27", medication: [{name: "Aspirin", dose: "14", unit: "tablets"}]}, 
{name: "Brian", age: "62", medication: []}, 
{name: "Chris", age: "46", medication: [{name: "Bisoprolol", dose: "3", unit: "ml"}, {name: "Clotrimazol", dose: "2", unit: "mg"}]}, 
{name: "David", age: "68", medication: [{name: "Aspirin", dose: "4", unit: "tablets"}, {name: "Bisoprolol", dose: "1.5", unit: "ml"}, {name: "Clotrimazol", dose: "2", unit: "mg"}]}
]

Now obviously, I can easily group those by, for example, age or name.

The problem arises when I want to show medications (by name), as any patient can have one, none, or several of those. Each patient can therefore be in none or several groups. So far, I see three ways of doing this:

duplicate patients with more than one medication, each copy with a different single medication

This creates redundancies, also I want draw a whole barrage of diagrams from the same data. The key for nesting (the function that returns the value to group by for each patient) will be created dynamically from a user interface - as it is a getter, not a setter, I would need to figure out how to reset medication for each copy, too. The copying might, depending on the complexity of my raw data, also not be trivial.

Concatenate medication names in the key function (with a unique separator) and split and add as needed when drawing.

Every occurring combination of medications will have its own group, and I would need to calculate sums from all groups containing a specific medication to get the total of it's frequency. This seems like a bad idea.

Write my own nest function that handles arrays.

I would lose d3's rollup and sort functionality unless I implement that, too, from scratch.

Is there a better way?

Nixie · Accepted Answer · 2016-09-24 18:23:23Z

1

I do not see a better way, except pre-processing your data (like your pt.1), so that the keys could be medication names, and values could be records from original data. You do not need to duplicate patients, you can just duplicate references to them, and that does not create a lot of redundancies (and saves memory). Instead of passing data into your charting function, you just pass extractMedications(data) instead, which is also not that really redundant.

function extractMedications(d)
{
    var res = [];
    d.forEach(function(patient) {
      patient.medication.forEach(function(medication) {
        res.push({medication_name: medication.name, patient: patient});
    })
  })
  return res;
}
console.log(extractMedications(data))

The other advantage that in this solution you can modify patient name in data array and that will be also reflected in the transformed array (because it holds a reference to the object from data)

Then you need to forward this structure to the d3.nest and do whatever you want.

answered Sep 24, 2016 at 18:23

Nixie

6374 silver badges9 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Klaus Hilgenfelder Over a year ago

This is an interesting idea, though I have since written my own nesting - as I work with data of which I don't know beforehand if it has arrays or how deep the nesting is going to be, it would have led to complicated checks and a completely different data structure for array data compared to "regular" data (or an unnecessary step to re-write data without arrays). A new nesting function has the added benefit that it saves me several passes over the full data, as I can do filtering and ordering within it.

Collectives™ on Stack Overflow

Nest data containing arrays with d3.nest

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related