In [1]:
import pandas as pd
import geopandas as gpd
import numpy as np
import os
import os
from scipy import stats
from sklearn import preprocessing
from sklearn.model_selection import KFold
from sklearn.linear_model import LinearRegression
import seaborn as sn
from scipy import stats
import matplotlib.pyplot as plt
%matplotlib inline
In [2]:
mitigationfile = pd.read_csv(r'C:\Users\i23733\Desktop\notebooks\projects\mitigation\distance.csv')
mitigation = pd.DataFrame(mitigationfile)
#mitigation = mitigation.fillna('')
c:\python27\lib\site-packages\IPython\core\interactiveshell.py:2714: DtypeWarning: Columns (48,49,50,51,52,53,55,56,58,59,62,88) have mixed types. Specify dtype option on import or set low_memory=False. interactivity=interactivity, compiler=compiler, result=result)
In [3]:
list(mitigation.columns.values.tolist())
Out[3]:
['wkt_geom', 'fid', 'Join_Count', 'TARGET_FID', 'OBJECTID', 'st_num', 'st_name', 'numocc', 'specneed', 'StructureT', 'struct_num', 'vis_add', 'gate', 'ing_egg', 'road_cond', 'BridgeWeig', 'DrivewayWi', 'drivelen', 'turn', 'roof', 'clean', 'eaves', 'Vents', 'bld_ext', 'decks', 'comb_mat', 'propane', 'site_water', 'fveg_z1', 'sveg_z1', 'lad_fl1', 'fveg_z2', 'sveg_z2', 'lad_fuel_2', 'contfuel', 'StructAlig', 'slope', 'setback', 'pos_slope', 'aspect', 'FFRiskAcce', 'FFRiskProp', 'FFRiskOHPo', 'FFRiskPets', 'FFRiskHazM', 'FFRiskEsca', 'FFRiskSola', 'notes', 'lat_str', 'long_str', 'own_name', 'taxlot', 'road_width', 'util', 'zone_1_mit', 'firesafe', 'home_mit', 'relhaz', 'hazclass', 'firedist', 'NewTaxLot', 'County', 'Prop_ID', 'FamDisPlan', 'PrepEmergK', 'IDEvacRout', 'RSGPartici', 'HomeCheckl', 'EvacDrill', 'RFI', 'FirstRespR', 'DateCollec', 'ParcelID', 'StructRisk', 'ParcelRisk', 'TotalRisk', 'HomeGoKit', 'RegisterdF', 'vegetation', 'ExtremeFFR', 'ExtremeF_1', 'GuessCoordinates', 'GuessedGeocode', 'lat', 'lon', 'BestAddress', 'City', 'State', 'Zip', 'FUEL', 'SLOPEfl', 'ACCESS', 'ADJ_FUEL', 'ADJ_SCORE', 'llFUEL', 'llSLOPE', 'llACCESS', 'llAFUEL', 'llWFHS', 'year_', 'acres', 'fire_name', 'loss', 'closestyear', 'closestfirename', 'closestdistance_miles', 'closestdistance_miles_']
In [49]:
mitigation['slope'].unique()
Out[49]:
array([ 4., nan, 2., 1., 3., 5.])
In [50]:
mitigation['SLOPEfl'] = mitigation['SLOPEfl'].replace({'S1':1,'S2':2,'S3':3,'S5':5})
In [20]:
# get rid of anything that doesn't have a FL score
mitigation = mitigation.loc[mitigation['ADJ_SCORE'] >= 0]
len(mitigation)
In [23]:
# Get the names of the fires where there were losses
# create a list to select the points in close prox to those fires
#loss = mitigation.loc[(mitigation.loss.notnull())]
loss = mitigation.loc[(mitigation.loss == 1)]
x = list(loss['closestfirename'].unique())
print x
['Moccasin Hill', 'BISCUIT COMPLEX', 'Chetco Bar']
In [24]:
def corr(df):
result = df.corr()
print sn.heatmap(result, annot=True, fmt = '.1g', cmap= 'coolwarm')
In [55]:
fields = ['SLOPEfl','slope','setback','pos_slope','aspect','StructRisk','ParcelRisk','TotalRisk']
fields2 = ['StructRisk','ParcelRisk','TotalRisk','roof','clean','eaves','Vents','ADJ_SCORE','ADJ_FUEL','relhaz']
In [56]:
# look at everything
mit = mitigation[fields]
#mit['loss'] = mit['loss'].fillna(0)
corr(mit)
AxesSubplot(0.125,0.125;0.62x0.755)
In [46]:
# looking at just the properties within 2 miles of the fires where there were damages
# losses & within 2mi - not damaged
prx = mitigation.loc[(mitigation.closestdistance_miles_ < 2) & (mitigation.closestfirename.isin(x))]
prx = prx[fields]
prx['loss'] = prx['loss'].fillna(0)
corr(prx)
AxesSubplot(0.125,0.125;0.62x0.755)
In [42]:
# just losses
loss = mitigation.loc[mitigation['loss'] == 1]
loss = loss[fields2]
corr(loss)
AxesSubplot(0.125,0.125;0.62x0.755)
In [43]:
# excluding losses & excluding those within 2mi of perimeter
outside2mi = mitigation.loc[(mitigation.closestdistance_miles_ > 2)]
outside2mi = outside2mi[fields2]
outside2mi['relhaz'] = outside2mi['relhaz'].dropna()
corr(outside2mi)
AxesSubplot(0.125,0.125;0.62x0.755)
In [33]:
len(outside2mi)
Out[33]:
21982
In [ ]:
# clean out '.'
#mitigation = mitigation.apply(lambda x: x.str.replace('.','') if x.name in ['roof','clean','eaves','Vents','bld_ext','decks'] else x, axis=1)
#mitigation['hazclass_numeric'] = mitigation['hazclass'].replace({'Negligible':1, 'Very Low':2,'Low':3, 'Moderate':4, 'High':5, 'Very High':6, 'Extreme':7})
In [ ]: