|
The Soft Neighborhood Model for Portland Public Schools
|
The
Soft Neighborhood Model is:
- A stable model for school assignment that dynamically balances class sizes
throughout the district, without boundaries and without boundary changes.
- A neighborhood model that assigns students to schools near their homes.
- An equitable model that encourages diverse classrooms and discourages socioeconomic
stratification.
We have prepared a 2-page overview:
We are working on having this overview translated into other languages. If you are fluent in another
language and would like to help, send us an email.
Our profuse thanks goes to Daodao Zhang for the translation into Chinese!
The most comprehensive document describing the Soft Neighborhood Model is our original report:
"The Soft Neighborhood Model:
A Dynamic Enrollment Balancing Framework"
This document is continually evolving.
Version 3.0 of this document revamps the experimental results on real PPS data, using
map-routed driving distances, computing additional metrics, and comparing several SNM configurations
to historical results.
Version 2.0 of this document includes a new section explaining how this model can be used
to handle acute enrollment problems (now Section 8, originally Section 6).
Our reference implementation of the Soft Neighborhood Model
was used to provide the results for the above paper. The code
for this will eventually be available here. In the meantime, here is the
historical PPS enrollment data which we used to produce those results:
If anyone has obtained any redistributable new/revised/corrected data sets from PPS, please let us know.
We would be happy to post them here and distribute them for you!
Please send your questions, comments, and suggestions to
feedback@softneighborhoodmodel.org.
Revisions:
- soft-neighborhoods-v3.1.pdf fixed a small rounding error
in computing the enrollment balancing metric (effectively rounding down all section sizes to the lower
of two values for 2008-15 cumulative runs), updated Figure 21 and Tables 11, 12 accordingly.
- soft-neighborhoods-v3.0.pdf results on PPS data overhauled,
expanded from one section into three:
- using routed driving distances instead of "crow's flight" cartesian distances
- using more accurate/relevant section counts for each school
- examined six configurations of Soft Neighborhood Model, as well as "single closest school"
- introduced several new metrics based on feedback from community meeting
- added diagrams showing sample SNM assignments and comparison to historic ones
- soft-neighborhoods-v2.0.pdf added new section on handling acute enrollment
problems
- soft-neighborhoods-v1.2.pdf clarify author email addresses
- soft-neighborhoods-v1.1.pdf updated with public comment text
- soft-neighborhoods-v1.0.pdf original release
Copyright 2015-2017, Brooke Cowan and Matthew Marjanovic