Data Dive 3.4

Where do LA's 311 requests really cluster?

A map made of six-sided tiles. Change their size and see what happens.

You will need paper and a pencil, and a browser for step 3

3.40 / 6Leave

Step 1 of 5 · Your guess

What do you think?

A city planner wants to build a new service depot where the most 311 requests happen.

She shows a map of Los Angeles cut into hexagons, six-sided tiles. Each tile is shaded by how many requests fell inside it. The darkest tile is in Westchester, so that is where the depot goes.

Write it down before you lookIf the tiles were bigger, would the darkest tile still be in Westchester? Write yes or no.
3.4 · Where do LA's 311 requests really cluster?1 / 6Leave

Step 2 of 5 · Look

The requests, cut into 2 km tiles

The same 1,500 requests, binned at 2 kmhexagon centres - only the bin size changes between these twolatitudelongitudea hexagon holding requeststhe busiest bin33.752433.881334.010134.13934.2679-118.583-118.48-118.377-118.273-118.17

Los Angeles 311 requests. The busiest tile is marked.

How to read it. Each hexagon is one tile, 2 kilometers across. The darker it is, the more requests fell inside it.

Find out
  1. Find the busiest tile. Where is it?
  2. Who decided how big the tiles should be?
3.4 · Where do LA's 311 requests really cluster?2 / 6Leave

Step 3 of 5 · Try it

Now you try it, on the same data

This opens the pool, the site's chart tool. The words in bold below are buttons and menus you will see on the screen.

Do these five things

Open the pool on this file: data.sfhsbta.com/thepool?file=dots_la_311_points.csv&mark=hexbin

  1. The map opens with hexagon tiles. Set tile size to 2 km. Hover the darkest tile and write down where it is.
  2. Set tile size to 5 km. Hover the darkest tile. Write it down.
  3. How far apart are your two answers? That gap is the finding.
  4. Try 3 km and 4 km. Does the answer ever settle down?
  5. Write one sentence the planner should print next to the map.
3.4 · Where do LA's 311 requests really cluster?3 / 6Leave

Step 4 of 5 · Say it

Your answer, in one sentence

I think the depot should go <where, or not yet> because <what the chart shows> and I checked it by <what you did to the chart>.

The last part matters most. It shows you did not just look at the chart, you tested it.

Here is one exampleI think the depot should not be placed from this map alone, because the darkest tile moved 15 km when I changed the tile size. I checked by setting tile size to 2 km and then 5 km and hovering the darkest tile each time.
3.4 · Where do LA's 311 requests really cluster?4 / 6Leave

Step 5 of 5 · What you found

Check your numbers

33.997 N, 118.433 W
the darkest tile at 2 km: Westchester
34.051 N, 118.283 W
the darkest tile at 5 km: Westlake
15 km
how far the answer moved
The big idea: the size of the zones decides the mapNothing about the requests changed. Only the tile size did, and the busiest spot moved across the city. Any map cut into zones, like districts or ZIP codes, has this problem. Whoever draws the zones can change the answer.
One more thing to think aboutCouncil districts are zones someone drew. What might dive 1.2's chart look like if the district lines were somewhere else?
3.4 · Where do LA's 311 requests really cluster?5 / 6Leave

After the dive

The charts you used today

the hexagonal binning map

When there are too many dots to read, this counts them into tiles so you can see where they pile up.

What it leaves outChange the size of the tiles and the hot spot can move. Same data, different answer.
the dot density map

One dot for a fixed number of things, dropped where they happen. Where the dots crowd together, so does the thing.

What it leaves outThe dots are scattered at random inside each area — they are not real addresses. A coarse dot-per-count can make a busy place look empty.

library · ladder · dives · the pool · this dive’s worksheet

3.46 / 6Leave