v1.0-reproduce

This release contains the data needed to reproduce the thesis results. The data was downloaded from Mapzen Metro Extracts on March 4, 2016 for Toronto, Ontario area.

1. Ensure that the required code installation is successful (see win or linux)
2. Download the source code (zip or tar) and data for this release
3. Edit py/config.txt to use the thesis data: src = 'path/to/toronto_canada.imposm-geojson.zip'
4. Use the console and navigate to the unpacked code folder: cd path/to/msa-thesis-1.0-reproduce/py
5. Run the code on the edited config.txt file: python thesis.py config.txt path/to/output_folder

The config.txt should be edited to change only path/to/toronto_canada.imposm-geojson.zip:

[settings]
cores = -1
[[plot]]
plot_style = 'whitegrid'
plot_dpi = 600
plot_ext = '.png'
[[analysis]]
cross_validation_tests = [2, 5, 10]
high_correlations = [-0.7, 0.7]
outlier_value = 10
persist = True
[[forest]]
n_estimators = [10, 64, 96, 128]
criterion = ['entropy']
class_weight = ['balanced']

[toronto_osm]
title = 'Toronto OpenStreetMap'
src = 'path/to/toronto_canada.imposm-geojson.zip'
filter = ['places', 'amenities', 'transport_areas', 'aeroways', 'transport_points', 'roads_gen1']
id = ['osm_id']
keep_columns = []
epsg = '26917'
units = 'meters'
target = 'type'

Running py/thesis.py:

cd path/to/msa-thesis-1.0-reproduce/py
python thesis.py config.txt path/to/output_folder