cweinschenk/fire-risk

Numerical Models for Developing Community Scale Risk Model

fire-risk

Numerical Models for Developing Community Scale Risk Model

Installation

Installing from source:

git clone https://github.com/FireCARES/fire-risk.git
cd fire-risk
pip install .

Running the tests

After installing from source, the doctests and unit tests can be executed from the fire-risk root directory using pytest:

py.test

Running the Differential In Standard Time model

from fire_risk.models import DIST

dist = DIST(floor_of_origin=34, beyond=12, object_of_origin=170, room_of_origin=190, building_of_origin=74, floor_extent=False)
dist.gibbs_sample()

Running the Differential In Standard Time model when residential fire counts are stored in a Postgres database

from fire_risk.models import DIST
from fire_risk.backends import PostgresBackend
from fire_risk.backends.queries import RESIDENTIAL_FIRES_BY_FDID_STATE

with PostgresBackend(dict(host='localhost')) as backend:
    # Get residential fire counts for Arlington, VA
    results = backend.get_firespread_counts(query=RESIDENTIAL_FIRES_BY_FDID_STATE, query_params=('01300', 'VA'))

dist = DIST(floor_extent=False, **results)

print dist.gibbs_sample()

Running the Differential In Standard Time model with an arrival time generated from a Lognormal

Fit of GIS routed arrival times.

import pandas
from fire_risk.models import DIST
from fire_risk.backends import PostgresBackend
from fire_risk.backends.queries import RESIDENTIAL_FIRES_BY_FDID_STATE
from fire_risk.utils import LogNormalDraw
from scipy.stats import lognorm

dt = pandas.read_csv('detroit_response_time.csv')

with PostgresBackend(dict(host='localhost')) as backend:
    # Get residential fire counts for Arlington, VA
    results = backend.get_firespread_counts(query=RESIDENTIAL_FIRES_BY_FDID_STATE, query_params=('01300', 'VA'))

    params = lognorm.fit(dt['total_travel_time'])
    # total_travel_time is in minutes use a 60 second multiplier
    results['arrival_time_draw'] = LogNormalDraw(*params, multiplier=60)

dist = DIST(floor_extent=False, **results)

print dist.gibbs_sample()