Unassigned

I decided to take a closer look at the 32 folks who were unassigned in the 2011 Kindergarten Lottery. While doing so, I also spotted some strange things that I have asked the FIC about. Specifically, I see two students that were pre-assigned to Barkstall and Bottenfield. Not Special Ed. Were they withheld? That’s fine if they were – parents decide to do that often enough if their child is on the really young end.

Anyway, those 32….

  • 7 did not choose Barkstall as their first choice (meaning the other 25 did); 4 chose Bottenfield first, 3 chose South Side first
  • Of the 25 that chose Barkstall first, 1 was Proximity A, 8 were Proximity B
  • 16 (or 23.1%) of those assigned to Barkstall were Proximity B
  • One of the non-proximity Barkstall unassigned choices has a “sibgroup” of 30. Siblings?
  • 16 of the unassigned (50%) filled in all 5 schools; among the rest, 4 filled in 4, 10 filled in 3, 2 filled in 2
  • Outside of the first choice, Barkstall was chosen 3 times (1 for 2nd, 1 for 3rd, 1 for 5th), Bottenfield was chosen 28 times (17 for 2nd, 5 for 3rd, 5 for 4th, 1 for 5th), South Side was chosen 15 times (3 for 2nd, 6 for 3rd, 3 for 4th, 3 for 5th)
  • One person (low-ses) was assigned to Barkstall as a 2nd choice. This was the only one who did not choose Barkstall first

Some of those stats are just plain weird; like the “sibgroup” one, especially if that indicates a sibling at the school. It is obvious that a majority of those that were unassigned had the misfortune of choosing “overchosen” schools (not just once, but many times). I see this as a failure due the complexity of the process and the lack of easily understood information.

Some (ongoing) thoughts:

  1. First and foremost, make the process more clear and intuitive, and spell out what “overchosen” means and which ones are such
  2. Exchange the keyword “Choice” with “Preference”
  3. Allow a parent to rank-order all schools
  4. If child is not assigned according to preference, assign (by default) to closest school available
  5. Bring it all in-house and put it online
PS – It may seem that #1 is mutually exclusive to #3, #4 and #5. What can I say? 🙂

2011 Kindergarten Lottery data made available

https://thecitizen4blog.files.wordpress.com/2011/04/champaign-kindergarten-lottery-document-schultz-foia-redacted.xls

Busy day today. 🙂

I have imported this into the online database, but I have not yet built up the php pages to query it. In the meantime, you can download the spreadsheet yourself and browse it. This could be like group-sourced analysis.

The data is now available in the database:

http://lottery.cb-pta.com/lottery/2011/

Some of the predefined queries do not work 100% because the data changed around a little, but most of them do. Enjoy.

A technical read – using TIGER/Line shapefiles and Google Maps

You might be wondering “what they heck does this have to do with Champaign schools?” Good question. 🙂 I tend to be a picture-oriented person, so let me draw your attention back to the Kindergarten overlap map:

http://lottery.cb-pta.com/other/cu-k-maps.html

There are several steps involved here. If you take a peek at the javascript (including the drawDistrictLine.js file), you will see that I am hardcoding all the geocodes for the school addresses and the school district boundary line. I am doing that for speed – otherwise, I have to resolve all those points when you load the page. I like speed.

The School District Boundary line

To get the school district boundary line, I started asking a bunch of questions (seems to be my modus operandi). I was not able to find anyone within Unit 4 that had the exact data points for the district boundary, so I started looking around http://www.census.gov/. There is a ton of crazy stuff out there and it completely blew me away. But for the most part, it is largely unintelligible, even for me. *grin* I poked around with some local resources: Tom Laue with the Champaign GIS Consortium, Karen Hogenboom of the UI’s Library Gov Information department, and Marilyn Ruiz of Pathobiology (not sure how Vet Med got such a huge interest in Spatial relationships, but hey). Ms. Hogenboom took the time to show me the ArcGIS system that the University has a license for. Wow, is that beast complicated. I tried to play with it for a little while, but ultimately gave up in light of easier solutions. What easier solutions, you say? Ms. Hogenboom also pointed me to the TIGER/Line shapefiles. Mumbo jumbo, I know. So I needed a way to convert these shapefiles to something I could read. Thanks (again) to Google, I came across two utilities, Shape2Text and shp2text. Don’t bother with the former (I did not even include a link for it). I like the second better for a couple different reasons; 1) it works, 2) the source code is included so I can see what it is actually doing. This utility will read the complex shapefile that you download from TIGER and spit out a human-readable text file. The format is a bit strange, but workable. I grabbed the shapefile for Illinois Unified School Districts, converted it with shp2text and stripped out the information that was pertinent to Unit 4. Oddly enough, the school districts is technically defined as two separate sections, which you can see on the map. What did I strip out? I grabbed all the geocode points that define the boundary. These are lat/long (latitude and longitude) pairs, for example 40.10620700 -88.35058900 (40 degrees north, 88 degrees east, or negative 88 degrees west). More about these geocodes in a second.

School circles

For the schools, I had several things to overcome. At first, I grabbed all the addresses and called a Google Map API function to resolve them to geocodes. However, I found that the page was taking a few microseconds to load them, not to mention that the API call is execute asncronously, thus if I want to attach them to the map with special identifying information, I have to somehow build a DOM array to keep track of the asynchronous calls. Big pain in the arse. So to massively simply things, I opted to make the geocodes static. The next part is actually drawing the circle. Fortunately, I found an example to follow. I love examples! So I altered the code from the example and drew 1.5 mile circles around the center, marked by the geocode for each school, with its appropriate marker. Next, I wanted a dynamic way to shade the circles. I came across a fading menu button example, which I adopted for my circles. I am not entirely happy with it, yet – you will see that there is a precedence if you mouse-over the circles, and some circles are really hard to light up due to the low precedence and high amount of overlap (especially South Side – there is only a small area that will allow you to see the brown circle).

Putting it all on the map

So, finally, adding the pieces together. Chronologically, I did the circles first, so I simply reused the circle idea to draw the district boundary line(s). Basically, you build an array of lat/long pairs and pass that into a polygon generator function that Google provides. Google also provides a polyline function, but polygons can be filled. 🙂 All the lat/long pairs are passed into the the polygon function to created that crazy, all-over-the-place district boundary. Next, I use some math (again, copied from the example) to generate an array of points – if you look closely at the circle (max zoom) and the source code, you will see that I only use 52 points to describe the circle, much like a 52-sided polygon. After generating the polygons, I have to add them to the map. Or, more accurately, pass the map into the polygon with the “setMap” function. I found that I had to add the district boundaries first, or else they would have higher precedence and you would not be able to mouse-over the circles.

I still want to pop up an info tab, much like what the fine folks at LocalSchoolDirectory have done. Their site is really slick, and I hope to make mine a little more fresh in terms of colors, shapes and general appeal. Maybe some day….. (/me has design envy). On the other hand, if you want that extra info, you can always go visit them. =)