torque/docs/guides/05-how-spatial-aggregation-works.md

How spatial aggregation works

When the Torque library renders points, it does not render exactly the same points you have in the database; instead it aggregates the points in clusters in order to speed up rendering and data transfer.

So imagine you have this CartoCSS:

Map {
  -torque-aggregation-function:"count(cartodb_id)";
  -torque-resolution: 2;
}

This means that for the current zoom level, Torque will fetch points in clusters of 2x2 pixels. Every cluster has a value calculated by the function defined after -torque-aggregation-function. In the case above, the value will be the number of points inside that cluster. That value can be accessed from CartoCSS using the variable value.

Every cluster is renderer as a point.

Given that you can do:

#layer {
[value > 1] { marker-fill: #000; }
[value > 4] { marker-fill: #400; }
[value > 16] { marker-fill: #800; }
[value > 32] { marker-fill: #F00; }
}

This would render the point with different colors depending on the number of points inside it.

Can I use strings with Torque?

In general you can not do:

[column = 'mytext'] { marker-fill: red; }

There are two reasons for this limitation:

• cluster does not contain values for all the columns, you can only use value variable
• you would need to use an aggregation function for strings

So how could I use strings column with Torque?

Imagine you have a string column (team) with two values, "team A" and "team B", and you want to color "team A" points to be red and "team B" to be blue, you could add a new column on the fly:

torqueLayer.setSQL("select *, (CASE WHEN team='team A' THEN 1 ELSE 2 END) as team_n from table");

and then apply this CartoCSS:

Map {
  ...
  -torque-aggregation-function: "round(avg(team_n))";
  ...
}

#layer {
  ...
  marker-fill: #FF0000;
  // avg of 1 and 2
  [value > 1.5] { marker-fill: #0000FF; }
  ...
}

Tip: CartoCSS is one the tools of the CARTO platform. You can learn more about the different components of CARTO platform, or directly dig into CartoCSS too details.