A framework to analyze traffic incident reports from the Bing Maps API and predict future incident levels

1. Install Pip
2. In the top folder run pip install -r requirements.txt

python pull_data.py [region1,region2,etc]

If no regions are specified only the sf data will be pulled by default

Simply run python export_arff.py [in_file.json] [out_file.arff] [conf_file.json]

An example configuration file can be found at conf/example.json. The maps and filters listed will be run in the order listed in your conf file.
Attributes have three fields:

1. The name of the attribute in the exported ARFF file,
2. The JSON path to the attribute in each instance
3. The type of attribute (NOMINAL, NUMERIC, and INTEGER are currently supported).

Note that since attributes will be extracted after your maps are run, you can add computed attributes in a map and then reference them in the attribute list. For example, in the example.json conf file the calculated attribute hour is referenced.