Herroin Overdose Analysis of Cincinnati, Ohio¶

MUSA 620 Final Project¶

Shuyan Hong & Jiazuo Zhang

1. Exploratory Analysis¶

import numpy as np
import pandas as pd
import geopandas as gpd
import folium
from matplotlib import pyplot as plt
import cartopy.crs as ccrs

Import overdose data.

#Overdose data
query = ("https://data.cincinnati-oh.gov/resource/7mtn-nnb5.json?$limit=10000")
raw_data = pd.read_json(query)
raw_data.head()

Firstly we map the overdose situation of of recent years using Folium.

df0 = raw_data[['latitude_x','longitude_x','incident_type_id']]
df0 = df0.dropna()

def label_race (row):
    return 'red';

df0['color'] = df0.apply (lambda row: label_race(row), axis=1)

# Folium map of Cincinnati
m1 = folium.Map(
    width=700,height=500,
    location=[39.15, -84.55],
    zoom_start=11,
    tiles="Stamen Toner"
)

cityBoundary = 'https://opendata.arcgis.com/datasets/ed78f4754b044ac5815d0a9efe9bb336_1.geojson'
folium.GeoJson(cityBoundary).add_to(m1)


# Heroin overdose points by numbers and incident type
import folium.plugins
#folium.plugins.MarkerCluster()
marker_cluster = folium.plugins.MarkerCluster().add_to(m1)

for i in range(0, len(df0)):
    folium.Marker(
      location=[ df0.iloc[i]['latitude_x'],df0.iloc[i]['longitude_x']],
      popup=df0.iloc[i]['incident_type_id'],
      icon=folium.Icon(color=df0.iloc[i]["color"])
   ).add_to(marker_cluster)
m1

from folium.plugins import HeatMap
# make a NumPy array (use the "values" attribute)
coordinates = df0[['latitude_x', 'longitude_x']].values

m2 = folium.Map(
    location=[39.15, -84.55],
    zoom_start=11,
    tiles="Stamen Toner"
)
HeatMap(coordinates).add_to(m2)
m2

Also we can create a hex bin map, and number of overdoses inside each hex bin.

data_valid = raw_data.loc[raw_data.latitude_x.notnull()]
data_valid = data_valid.loc[data_valid.longitude_x.notnull()]
#ignore null longs or lats

from datashader.utils import lnglat_to_meters as webm
data_valid.loc[:, "longitude_x"], data_valid.loc[:, "latitude_x"], = webm(data_valid["longitude_x"],data_valid["latitude_x"])
#change lon, lat to x,y

#bound = gpd.read_file(cityBoundary)
#bound.crs = {'init': 'epsg:3857'}

crs = ccrs.epsg('3857') # web mercator
ax = plt.axes()

# convert to Web Mercator and plot the hexbins 
#X = data_valid.to_crs(epsg=3857)
vals = ax.hexbin(data_valid.longitude_x, data_valid.latitude_x, gridsize=50, mincnt=1, cmap='coolwarm')

ax.set_aspect('equal')

# larger figure
ax.figure.set_size_inches((12, 12))

# add a legend
#and make colorbar fit the size of figure
from mpl_toolkits.axes_grid1 import make_axes_locatable
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.05)
cbar = plt.colorbar(vals,cax=cax)
cbar.ax.tick_params(labelsize=10)

2. Regression Analysis¶

By exploring different possible factors that might have impact on numbers of overdoses of each tract, we run the linear regression model to verify this.

2.1 Predictors¶

#!pip install cenpy
#!pip install pysal
import cenpy as cen
import pysal
import requests

/Users/zhang/anaconda3/envs/musa620/lib/python3.6/site-packages/pysal/model/spvcm/abstracts.py:10: UserWarning: The `dill` module is required to use the sqlite backend fully.
  from .sqlite import head_to_sql, start_sql

Now we read data of some predictors. The data includes 311 requests, black and white people percentage, median income, balchelor percentage, crime level, disatance to vanant properties and distances to transit stations and hospitals.

We use census package to acquire census tract data of Cincinnati, instructions from https://github.com/datamade/census

#!pip install us
from census import Census
from us import states
MY_API_KEY = '97e85eda8a09ad9759d9cbc91a8d3f9eb71b9c0c'
c = Census(MY_API_KEY)

# population of all tracts of Hamilton County, from acs5 
population_raw = c.acs5.state_county_tract('B01003_001E', states.OH.fips, '061', Census.ALL)
population = pd.DataFrame(population_raw)
population = population.rename({'B01003_001E':'population'}, axis=1)

#  bachelor number of all tracts of Hamilton County, from acs5 
bachelor_raw = c.acs5.state_county_tract('B15003_022E', states.OH.fips, '061', Census.ALL)
bachelor = pd.DataFrame(bachelor_raw)
bachelor = bachelor.rename({'B15003_022E':'bachelor'}, axis=1)

#  black population of all tracts of Hamilton County, from acs5 
black_pop_raw = c.acs5.state_county_tract('B02001_003E', states.OH.fips, '061', Census.ALL)
black_pop = pd.DataFrame(black_pop_raw)
black_pop = black_pop.rename({'B02001_003E':'black_pop'}, axis=1)

#  white population of all tracts of Hamilton County, from acs5 
white_pop_raw = c.acs5.state_county_tract('B02001_002E', states.OH.fips, '061', Census.ALL)
white_pop = pd.DataFrame(white_pop_raw)
white_pop = white_pop.rename({'B02001_002E':'white_pop'}, axis=1)

#  median household income of all tracts of Hamilton County, from acs5 
median_hh_raw = c.acs5.state_county_tract('B19013_001E', states.OH.fips, '061', Census.ALL)
median_income = pd.DataFrame(median_hh_raw)
median_income = median_income.rename({'B19013_001E':'median_income'}, axis=1)

df1 = pd.merge(population, bachelor, on='tract', how='outer')
df2 = pd.merge(df1, black_pop, on='tract', how='outer')
df3 = pd.merge(df2, white_pop, on='tract', how='outer')
df4 = pd.merge(df3, median_income, on='tract', how='outer')

# calculte black and white percetage
demographic = df4[['tract','population','bachelor','black_pop','white_pop','median_income']]
demographic['black_pct'] = demographic['black_pop']/demographic['population']
demographic['white_pct'] = demographic['white_pop']/demographic['population']
demographic['bachelor_pct'] = demographic['bachelor']/demographic['population']

/Users/zhang/anaconda3/envs/musa620/lib/python3.6/site-packages/ipykernel_launcher.py:3: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
  This is separate from the ipykernel package so we can avoid doing imports until
/Users/zhang/anaconda3/envs/musa620/lib/python3.6/site-packages/ipykernel_launcher.py:4: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
  after removing the cwd from sys.path.
/Users/zhang/anaconda3/envs/musa620/lib/python3.6/site-packages/ipykernel_launcher.py:5: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
  """

We want to see how many ovdedose records are inside each census tract these years. The census tracts data contains tracts that beyond the city limit, so we only choose tracts within the city limit.

# intersect city boundary and all census tracts to get census tracts within city boundary
boundary = gpd.read_file(cityBoundary)
cts_raw = gpd.read_file('https://opendata.arcgis.com/datasets/d02e89ac4bfd47caa46c93a056dcbab9_0.geojson')
cts = gpd.overlay(boundary, cts_raw, how='intersection')

# select demographic data of census tracts within city boundary
cts = cts[['TRACTCE10','geometry']]
cts = cts.rename({'TRACTCE10':'tract'}, axis=1)
cts_demographic = pd.merge(cts, demographic, on='tract', how='inner')

cts_demographic.head()

Then we calculate number of hospitals and number of schools.

cts1 = cts_demographic

# read hospital geojson
hospitals = gpd.read_file('https://opendata.arcgis.com/datasets/0baa0cdf409e46e1800a6f08081bd9f7_1.geojson')
hospitals.crs = {'init' :'epsg:4326'}

# read school geojson
schools = gpd.read_file('https://opendata.arcgis.com/datasets/97194decd96d451cb9f67c19078b80c9_6.geojson')
schools.crs = {'init' :'epsg:4326'}

iv_dfsjoin1 = gpd.sjoin(cts1,hospitals)# spatil join census tracts and hospitals
hospital_count = iv_dfsjoin1.groupby('tract').count()[['GLOBALID']]
hospital_count = hospital_count.rename({'GLOBALID':'hospital_count'}, axis=1)

iv_dfsjoin2 = gpd.sjoin(cts1,schools)# spatil join census tracts and schools
school_count = iv_dfsjoin2.groupby('tract').count()[['GLOBALID']]
school_count = school_count.rename({'GLOBALID':'school_count'}, axis=1)

2.2 Dependent Variable¶

new_cts = cts_demographic

#create geometry from df1
from shapely.geometry import Point
points = gpd.GeoDataFrame(df0, geometry=[Point(x, y) for x, y in zip(df0.longitude_x, df0.latitude_x)])

points.head()

points.crs = {'init' :'epsg:4326'}
dv_dfsjoin = gpd.sjoin(new_cts,points)# spatil join census tracts and overdose points

# count overdose points inside each tract
od_count = dv_dfsjoin.groupby('tract').count()[['index_right']]
od_count = od_count.rename({'index_right':'od_count'}, axis=1)
od_count.head()

Now merge all dependent and independent variables, and calculate overdose rate by od_count divided by population inside each tract.

merge1 = pd.merge(cts_demographic, hospital_count, on='tract', how='outer')
merge2 = pd.merge(merge1, school_count, on='tract', how='outer')
final_df = pd.merge(merge2, od_count, on='tract', how='outer')
final_df['od_rate'] = final_df['od_count']/final_df['population']
final_df = final_df.fillna(0)

final_df

2.3 Regression¶

Firstly we can create some interactive maps showing predictors and overdose rates using hvplot.

import hvplot.pandas

Now we do Linear Regression.

First we want to see the relationship of single predictors and the overdose rate.

Then we do the multivariable linear regression. We start by looking at the correlation matrix of predictors.

It is not surprising that median income, percent of whites and percent of bachelors are strongly assoviated.

As we can see from the summary, the overdose rate is positively related with percent of blacks, percent of whites, number of hospitals and number of schools, but negatvely associated wth median household income.

3. Clustering Analysis¶

Looking at how overdoses of different tracts in nearby neighborhoods have spatial relationships, whether they are clustered, or disparted, or randomly distributed.

We will create maps not only showing the clustering facts, but also give them interactive surfaces to generate detailed information of selected tract, such as population, median household income and bechalor number.

import altair as alt
from vega_datasets import data
alt.renderers.enable("notebook")

RendererRegistry.enable('notebook')

from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()

DBSCAN¶

from sklearn.cluster import dbscan

data2 = raw_data[['create_time_incident','latitude_x', 'longitude_x']]
data2 = data2.dropna()

# some parameters to start with
eps = 200 # in meters, maximum distance between two samples for one to be considered as in the neighborhood of the other
min_samples = 5 # number of samples (or total weight) in a neighborhood for a point to be considered as a core point

from datashader.utils import lnglat_to_meters as webm
data2.loc[:, "longitude_x"], data2.loc[:, "latitude_x"], = webm(data2["longitude_x"],data2["latitude_x"])
scaler = StandardScaler()
scaled_features = scaler.fit_transform(data2[['latitude_x', 'longitude_x']])

cores, labels = dbscan(data2[['longitude_x', 'latitude_x']], 
                       eps=eps, min_samples=min_samples)

# add our labels to the original data
data2['label'] = labels
# The number of clusters is the number of unique labels minus one (because noise has a label of -1)
num_clusters = data2['label'].nunique() - 1
print("number of clusters = ", num_clusters)

number of clusters =  140

print(len(cores))

4593

# Setup figure and axis
f, ax = plt.subplots(1, figsize=(10, 7), facecolor='black')

# Plot the noise samples in grey
noise = data2.loc[data2['label']==-1]
ax.scatter(noise['longitude_x'], noise['latitude_x'], c='grey', s=5, linewidth=0)

# loop over each cluster number
for label_num in range(num_clusters):
    
    # extract the samples with this label number
    this_cluster = data2.loc[data2['label'] == label_num]
    
    # plot the centroids
    ax.scatter(this_cluster['longitude_x'], this_cluster['latitude_x'], 
                linewidth=0, color='red')

# format and show
ax.set_axis_off()
plt.show()

N = data2.groupby('label').size()


# sort from largest to smallest
N = N.sort_values(ascending=False)
print(N)

# extract labels (ignoring label -1 for noise)
N2 = N.drop(labels=[-1])
top5 = list(N2.iloc[0:5].index)
print(top5)

label
 0      1189
-1      1005
 4       984
 9       204
 36      197
 45      153
 5       125
 13      104
 30       75
 20       74
 8        71
 3        70
 2        64
 39       62
 22       58
 33       47
 92       46
 48       36
 67       34
 96       33
 21       33
 47       32
 1        32
 10       30
 11       29
 101      28
 38       28
 123      27
 19       27
 98       24
        ... 
 68        6
 106       6
 66        6
 132       6
 100       6
 107       6
 94        6
 112       6
 133       5
 131       5
 128       5
 135       5
 136       5
 130       5
 97        5
 126       5
 25        5
 120       5
 118       5
 111       5
 109       5
 99        5
 54        5
 55        5
 138       5
 78        5
 85        5
 102       5
 139       5
 119       4
Length: 141, dtype: int64
[0, 4, 9, 36, 45]

4. Time-spatial Analysis¶

Looking at how overdose of each tract changes over the last three years (from 2016 to 2018) by creating a GIF image highlighting the concentrated tracts. The methods refer to Homework 5.

4.1 Overdose by months¶

import zipfile
import os
import urllib
import dask.dataframe as dd
import datashader as ds
import datashader.transfer_functions as tf

from datashader.colors import Greys9, inferno

# extract occur hour
data2['create_time_incident'] = pd.to_datetime(data2['create_time_incident'])
data2['month'] = data2.create_time_incident.dt.month

4.2 Overdose by years¶

data3 = data2
data3['create_time_incident'] = pd.to_datetime(data3['create_time_incident'])
data3['year'] = data3.create_time_incident.dt.year

4.3 Overdose trend¶

We can make plot of overdose counts by months and date.

data4 = data3.groupby(['year','month']).count()[['create_time_incident']]

data4 = data4.rename({'create_time_incident':'count'}, axis=1)
data4 = data4.reset_index()
data4

data4.hvplot(x='month',groupby='year',kind='bar')

data5 = data3
data5['date'] = data3.create_time_incident.dt.date
data5 = data5.groupby(['year','month','date']).count()[['create_time_incident']]
data5 = data5.rename({'create_time_incident':'count'}, axis=1)
data5 = data5.reset_index()
data5.hvplot(x='date',y='count',kind='bar')

5. The Online Visualization¶

import numpy as np
import pandas as pd
import geopandas as gpd
import requests

import holoviews as hv
import param as pm
import panel as pn

import altair as alt
import folium
from folium.plugins import HeatMap

hv.extension('bokeh')
pn.extension('vega')
alt.renderers.enable('notebook');

MAX = 12
MIN = 1

# store the start/end dates of our data set
MIN_DATE = data2['month'].min()
MAX_DATE = data2['month'].max()

# store the default min/max date
DEFAULT_BOUNDS = (MIN_DATE, MAX_DATE)

data6 = raw_data[['create_time_incident','latitude_x', 'longitude_x']]
data6 = data6.dropna()
data6['create_time_incident'] = pd.to_datetime(data6['create_time_incident'])
data6['month'] = data3.create_time_incident.dt.month

class CincinnatiOverdoseApp(pm.Parameterized):
    """
    An app visualizing recent overdoses in Cincinnati. 
    
    It includes:
    - a Folium heatmap
    - a map showing overdose rates and demographic data
    - a Hvplot line chart

    """
    
    # the number of days to get data for
    month = pm.Integer(default=7, bounds=(MIN, MAX))
    
    # the x-axis selection on the daily shootings chart
    timeFilter = hv.streams.BoundsX(boundsx=DEFAULT_BOUNDS)
    
    
    def filter_by_month(self):
        """
        Return the subset of the full data set ('data2') that 
        occurred in the selected month.
        """
        # select data based on month
        selection = data6[data6['month'] == self.month]  

        # only return subset of data that is necessary
        return selection
    
    @pm.depends("month", watch=True)
    def _reset_timeFilter(self):
        """
        Internal function that resets the time filter to the default limits
        """
        self.timeFilter.update(boundsx=DEFAULT_BOUNDS)
    

    @pm.depends("month")
    def heatmap(self):
        """
        Return a Folium map with a heatmap showing the currently 
        selected data.
        """
        # initialize the Folium map
        m3 = folium.Map(
                location=[39.15, -84.55],
                zoom_start=11,
                tiles="Stamen Toner"
        )

        # get the filtered data
        data = self.filter_by_month()

        # convert to coordinates array
        coordinates = data[['latitude_x', 'longitude_x']].values

        # add heat map
        folium.GeoJson(cityBoundary).add_to(m3)
        HeatMap(coordinates).add_to(m3)
        # IMPORTANT: add map to a folium Figure
        # return a figure with a set width/height
        figure = folium.Figure(width=700, height=400)
        m3.add_to(figure)

        # return the Pane object
        return pn.Pane(figure)
    
    
    def hvplotmap(self):
        figure = final_df.hvplot(c='od_rate', width=700, height=400, dynamic=False,  
                                 cmap='coolwarm' , title = 'Overdose Rate (2015-2018)')

        # return the Pane object
        return pn.Pane(figure)
    
    def barplot(self):
        figure = data5.hvplot(x='date',y='count',kind='line',width = 1400,title = 'Overdose Count by date')

        # return the Pane object
        return pn.Pane(figure)

app = CincinnatiOverdoseApp(name="")

# Create some HTML elements
title = pn.Pane("<h2>Heroin Overdose in Cincinnati</h2>", width=500)
instructions = pn.Pane(
    """
<div font-size=28px><b>Note:</b> From he summer of 2015 to the summer of 2018, there are more than
6000 heroin overdose reports in Cincinnati, Ohio. This page shows situation and trend of heroin overdose in that city.</div>""",
    width=500,
)

panel = pn.Column(
    pn.Row(pn.Column(pn.Row(title))),
    pn.Row(pn.Row(app.param),instructions, align="center"),
    pn.Row(app.heatmap,app.hvplotmap, align="center"),
    pn.Row(app.barplot)
)

panel.servable()

	address_x	agency	arrival_time_primary_unit	beat	closed_time_incident	community_council_neighborhood	create_time_incident	dispatch_time_primary_unit	disposition_text	district	event_number	incident_type_desc	incident_type_id	latitude_x	longitude_x	priority	priority_color
0	W 6TH ST / RACE ST	CPD	NaN	P011	2018-08-18T23:13:00.000	DOWNTOWN	2018-08-18T23:04:31.000	NaN	HBF: HANDLED BY FIRE	1.0	CPD180818001582	NaN	HEROINP-COMBINED	39.101675	-84.514038	2.0	NaN
1	8XX POPLAR ST	CPD	2018-08-18T20:24:30.000	P131	2018-08-18T20:36:04.000	WEST END	2018-08-18T20:19:46.000	2018-08-18T20:21:50.000	HBF: HANDLED BY FIRE	1.0	CPD180818001335	NaN	HEROINP-COMBINED	39.115051	-84.529110	2.0	NaN
2	47XX KENARD AV	CPD	2018-08-18T19:03:33.000	P531	2018-08-18T21:48:01.000	SPRING GROVE VILLAGE	2018-08-18T19:03:33.000	2018-08-18T19:03:33.000	ARR: ARREST	5.0	CPD180818001232	NaN	HEROINP-COMBINED	39.164528	-84.513777	2.0	NaN
3	4XX VOLKERT PL	CPD	NaN	P511	2018-08-18T18:18:34.000	CUF	2018-08-18T18:00:54.000	NaN	HBF: HANDLED BY FIRE	5.0	CPD180818001142	NaN	HEROINP-COMBINED	39.126231	-84.525635	2.0	NaN
4	42XX FERGUS ST	CPD	2018-08-18T16:22:25.000	P531	2018-08-18T16:55:58.000	NORTHSIDE	2018-08-18T16:20:51.000	2018-08-18T16:22:18.000	AST: ASSIST	5.0	CPD180818000986	NaN	HEROINP-COMBINED	39.164508	-84.535824	2.0	NaN

	tract	geometry	population	bachelor	black_pop	white_pop	median_income	black_pct	white_pct	bachelor_pct
0	025104	(POLYGON ((-84.41357452291133 39.0585997697975...	1862.0	585.0	19.0	1749.0	166250.0	0.010204	0.939313	0.314178
1	025002	(POLYGON ((-84.36969552616698 39.0737372659736...	6823.0	1539.0	136.0	6495.0	88979.0	0.019933	0.951927	0.225561
2	004500	(POLYGON ((-84.40989906360086 39.0833943746703...	1034.0	170.0	12.0	997.0	92167.0	0.011605	0.964217	0.164410
3	004603	POLYGON ((-84.37303856194467 39.08145590495828...	3019.0	706.0	84.0	2746.0	66556.0	0.027824	0.909573	0.233852
4	004602	(POLYGON ((-84.37086892135163 39.0859844464269...	4832.0	1039.0	200.0	3919.0	49063.0	0.041391	0.811051	0.215025

	latitude_x	longitude_x	incident_type_id	color	geometry
0	39.101675	-84.514038	HEROINP-COMBINED	red	POINT (-84.514038 39.1016750001196)
1	39.115051	-84.529110	HEROINP-COMBINED	red	POINT (-84.52911 39.1150510001195)
2	39.164528	-84.513777	HEROINP-COMBINED	red	POINT (-84.513777 39.1645280001195)
3	39.126231	-84.525635	HEROINP-COMBINED	red	POINT (-84.52563499999999 39.1262310001196)
4	39.164508	-84.535824	HEROINP-COMBINED	red	POINT (-84.53582400000001 39.1645080001195)

	od_count
tract
000200	18
000700	268
000900	145
001000	55
001100	86

	tract	geometry	population	bachelor	black_pop	white_pop	median_income	black_pct	white_pct	bachelor_pct	hospital_count	school_count	od_count	od_rate
0	025104	(POLYGON ((-84.41357452291133 39.0585997697975...	1862.0	585.0	19.0	1749.0	166250.0	0.010204	0.939313	0.314178	0.0	0.0	0.0	0.000000
1	025002	(POLYGON ((-84.36969552616698 39.0737372659736...	6823.0	1539.0	136.0	6495.0	88979.0	0.019933	0.951927	0.225561	0.0	0.0	0.0	0.000000
2	004500	(POLYGON ((-84.40989906360086 39.0833943746703...	1034.0	170.0	12.0	997.0	92167.0	0.011605	0.964217	0.164410	0.0	0.0	4.0	0.003868
3	004603	POLYGON ((-84.37303856194467 39.08145590495828...	3019.0	706.0	84.0	2746.0	66556.0	0.027824	0.909573	0.233852	0.0	1.0	6.0	0.001987
4	004602	(POLYGON ((-84.37086892135163 39.0859844464269...	4832.0	1039.0	200.0	3919.0	49063.0	0.041391	0.811051	0.215025	0.0	3.0	45.0	0.009313
5	025001	(POLYGON ((-84.36957073546399 39.0749964865464...	6260.0	1548.0	62.0	6028.0	82868.0	0.009904	0.962939	0.247284	0.0	0.0	0.0	0.000000
6	004605	(POLYGON ((-84.41169303481355 39.1008944955807...	2516.0	492.0	237.0	2095.0	58650.0	0.094197	0.832671	0.195548	0.0	0.0	22.0	0.008744
7	004604	POLYGON ((-84.39448364121984 39.10524757245881...	3887.0	481.0	288.0	3329.0	51431.0	0.074093	0.856445	0.123746	0.0	0.0	62.0	0.015951
8	004702	POLYGON ((-84.40476674756641 39.13493087559406...	780.0	83.0	13.0	752.0	38646.0	0.016667	0.964103	0.106410	0.0	0.0	57.0	0.073077
9	026600	POLYGON ((-84.46571383669516 39.12193007521013...	1608.0	432.0	141.0	1404.0	70250.0	0.087687	0.873134	0.268657	0.0	1.0	82.0	0.050995
10	004701	POLYGON ((-84.42808849604683 39.12703657693825...	3196.0	1101.0	53.0	3077.0	95488.0	0.016583	0.962766	0.344493	0.0	2.0	6.0	0.001877
11	004200	POLYGON ((-84.46444366483048 39.12983595709321...	1836.0	518.0	424.0	1300.0	53056.0	0.230937	0.708061	0.282135	0.0	1.0	11.0	0.005991
12	004900	POLYGON ((-84.4389178289271 39.11773602336285,...	6487.0	1859.0	269.0	5818.0	94868.0	0.041468	0.896871	0.286573	0.0	2.0	13.0	0.002004
13	024901	(POLYGON ((-84.3970721740594 39.12196931007487...	1075.0	122.0	0.0	1065.0	57500.0	0.000000	0.990698	0.113488	0.0	0.0	0.0	0.000000
14	004800	POLYGON ((-84.40569287293694 39.13825103319869...	3426.0	1054.0	9.0	3196.0	119470.0	0.002627	0.932866	0.307647	0.0	1.0	3.0	0.000876
15	004100	POLYGON ((-84.45843631676497 39.1343519865754,...	1615.0	381.0	728.0	800.0	59643.0	0.450774	0.495356	0.235913	0.0	4.0	9.0	0.005573
16	003900	POLYGON ((-84.4724115280404 39.13872552367319,...	1951.0	108.0	1593.0	350.0	22566.0	0.816504	0.179395	0.055356	0.0	0.0	9.0	0.004613
17	005100	POLYGON ((-84.40851084594802 39.14021088964427...	2410.0	626.0	79.0	2258.0	94145.0	0.032780	0.936929	0.259751	0.0	1.0	3.0	0.001245
18	004000	POLYGON ((-84.46725090467903 39.14695878206831...	2372.0	320.0	1546.0	648.0	42949.0	0.651771	0.273187	0.134907	0.0	1.0	10.0	0.004216
19	024700	(POLYGON ((-84.40220717113631 39.1347695617024...	1707.0	263.0	59.0	1615.0	52297.0	0.034564	0.946104	0.154071	0.0	0.0	0.0	0.000000
20	003800	POLYGON ((-84.47189734296572 39.14697704190844...	1940.0	140.0	1753.0	181.0	22788.0	0.903608	0.093299	0.072165	0.0	2.0	26.0	0.013402
21	005000	(POLYGON ((-84.45444775941291 39.1459729946670...	5049.0	1406.0	339.0	4227.0	69000.0	0.067142	0.837195	0.278471	0.0	3.0	13.0	0.002575
22	024800	POLYGON ((-84.38558243035754 39.14844630157257...	3404.0	907.0	5.0	3234.0	96106.0	0.001469	0.950059	0.266451	0.0	0.0	0.0	0.000000
23	005302	POLYGON ((-84.40451981774257 39.14973722544627...	3249.0	923.0	304.0	2720.0	66250.0	0.093567	0.837181	0.284087	0.0	0.0	11.0	0.003386
24	005301	POLYGON ((-84.42516472407397 39.15575951288675...	3587.0	1071.0	475.0	2974.0	56563.0	0.132423	0.829105	0.298578	0.0	2.0	10.0	0.002788
25	025402	(POLYGON ((-84.44034958036744 39.1557611335698...	3052.0	580.0	265.0	2693.0	60741.0	0.086828	0.882372	0.190039	0.0	0.0	0.0	0.000000
26	005200	POLYGON ((-84.44279399384786 39.14528821560646...	3535.0	1084.0	196.0	3101.0	69318.0	0.055446	0.877228	0.306648	0.0	1.0	10.0	0.002829
27	025600	(POLYGON ((-84.45944560213331 39.1464920393789...	3523.0	574.0	311.0	3129.0	44500.0	0.088277	0.888163	0.162929	0.0	0.0	0.0	0.000000
28	005600	POLYGON ((-84.37905156837199 39.16008944046915...	5790.0	951.0	1928.0	3575.0	46105.0	0.332988	0.617444	0.164249	0.0	2.0	20.0	0.003454
29	025401	POLYGON ((-84.43934752457294 39.15653210139998...	1986.0	385.0	129.0	1824.0	43402.0	0.064955	0.918429	0.193857	0.0	0.0	0.0	0.000000
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
120	020901	(POLYGON ((-84.62305139318956 39.1459676064429...	3539.0	298.0	299.0	3020.0	45556.0	0.084487	0.853348	0.084205	0.0	0.0	2.0	0.000565
121	006500	POLYGON ((-84.46604105823579 39.16735990757773...	5894.0	445.0	2953.0	2764.0	46462.0	0.501018	0.468951	0.075501	0.0	2.0	54.0	0.009162
122	008502	POLYGON ((-84.55265169991918 39.16421849726655...	2094.0	30.0	1819.0	143.0	7044.0	0.868672	0.068290	0.014327	0.0	1.0	7.0	0.003343
123	020902	(POLYGON ((-84.60275325565836 39.1618617903557...	5548.0	603.0	876.0	4548.0	40176.0	0.157895	0.819755	0.108688	0.0	0.0	0.0	0.000000
124	010100	POLYGON ((-84.60815706077847 39.15113837434231...	5359.0	559.0	2946.0	2185.0	37884.0	0.549729	0.407725	0.104311	0.0	1.0	80.0	0.014928
125	007800	POLYGON ((-84.54501968760005 39.16714455545117...	2538.0	459.0	757.0	1639.0	38571.0	0.298266	0.645784	0.180851	0.0	2.0	58.0	0.022853
126	007400	POLYGON ((-84.52894142414301 39.16157064925422...	1716.0	373.0	281.0	1350.0	48750.0	0.163753	0.786713	0.217366	0.0	0.0	49.0	0.028555
127	007900	POLYGON ((-84.54346057775746 39.17476733751921...	1697.0	285.0	819.0	848.0	33083.0	0.482616	0.499705	0.167943	0.0	0.0	29.0	0.017089
128	025800	(POLYGON ((-84.48322645115884 39.1687728134858...	4366.0	421.0	773.0	3406.0	51029.0	0.177050	0.780119	0.096427	0.0	0.0	0.0	0.000000
129	020802	(POLYGON ((-84.58681466060455 39.1642494579820...	5310.0	970.0	461.0	4596.0	67083.0	0.086817	0.865537	0.182674	0.0	0.0	0.0	0.000000
130	008501	POLYGON ((-84.58258361702951 39.16988658576086...	3061.0	163.0	2202.0	773.0	18326.0	0.719373	0.252532	0.053251	0.0	0.0	43.0	0.014048
131	007500	POLYGON ((-84.54216972568472 39.18434683094865...	2121.0	359.0	498.0	1557.0	47647.0	0.234795	0.734088	0.169260	0.0	1.0	13.0	0.006129
132	006400	POLYGON ((-84.48527940902325 39.1727951596445,...	3382.0	272.0	3112.0	195.0	30748.0	0.920166	0.057658	0.080426	0.0	4.0	29.0	0.008575
133	007300	POLYGON ((-84.50562540689216 39.17536341551688...	1967.0	208.0	964.0	918.0	43450.0	0.490086	0.466701	0.105745	0.0	3.0	83.0	0.042196
134	020811	(POLYGON ((-84.57191468719508 39.1851453068084...	6826.0	840.0	1867.0	4603.0	64904.0	0.273513	0.674333	0.123059	0.0	0.0	11.0	0.001611
135	025700	(POLYGON ((-84.48190898514163 39.1888039171248...	1965.0	66.0	373.0	1405.0	30089.0	0.189822	0.715013	0.033588	0.0	0.0	0.0	0.000000
136	008400	POLYGON ((-84.54611722737476 39.18881689934389...	2187.0	196.0	1326.0	812.0	31875.0	0.606310	0.371285	0.089620	0.0	2.0	15.0	0.006859
137	008000	POLYGON ((-84.52089703660116 39.20122255647082...	5204.0	157.0	4260.0	643.0	11600.0	0.818601	0.123559	0.030169	0.0	1.0	21.0	0.004035
138	008100	POLYGON ((-84.53468716564535 39.20223882375489...	2752.0	210.0	2119.0	567.0	30526.0	0.769985	0.206032	0.076308	0.0	1.0	14.0	0.005087
139	006100	POLYGON ((-84.49410091979443 39.20130140101698...	2567.0	137.0	564.0	1555.0	31726.0	0.219712	0.605765	0.053370	0.0	1.0	153.0	0.059603
140	008202	POLYGON ((-84.55883515146103 39.20364065673207...	3279.0	474.0	2087.0	1120.0	44635.0	0.636475	0.341568	0.144556	0.0	3.0	20.0	0.006099
141	008300	POLYGON ((-84.58037897116463 39.19253946427716...	4178.0	429.0	2323.0	1699.0	45969.0	0.556008	0.406654	0.102681	2.0	1.0	52.0	0.012446
142	008201	POLYGON ((-84.55763442709959 39.20357901006025...	4095.0	479.0	2472.0	1237.0	42595.0	0.603663	0.302076	0.116972	0.0	2.0	18.0	0.004396
143	021900	(POLYGON ((-84.55760923423578 39.2038488125994...	1583.0	117.0	1054.0	523.0	33846.0	0.665824	0.330385	0.073910	0.0	0.0	0.0	0.000000
144	022200	(POLYGON ((-84.47438418424967 39.2098562541126...	4481.0	857.0	1110.0	3320.0	56321.0	0.247713	0.740906	0.191252	0.0	0.0	0.0	0.000000
145	011100	(POLYGON ((-84.54135362260716 39.2126544630221...	3357.0	432.0	1716.0	1505.0	56620.0	0.511171	0.448317	0.128686	0.0	0.0	7.0	0.002085
146	021802	(POLYGON ((-84.55104971477819 39.2093521842509...	2917.0	293.0	1273.0	1369.0	51071.0	0.436407	0.469318	0.100446	0.0	0.0	0.0	0.000000
147	021801	(POLYGON ((-84.53274294648506 39.2169644639702...	6405.0	676.0	3476.0	2764.0	42371.0	0.542701	0.431538	0.105543	0.0	0.0	0.0	0.000000
148	022102	POLYGON ((-84.53107073060511 39.21680448689365...	6626.0	871.0	2120.0	3656.0	64831.0	0.319952	0.551766	0.131452	0.0	0.0	0.0	0.000000
149	022601	(POLYGON ((-84.47363437549781 39.2176530190985...	6170.0	1191.0	664.0	4995.0	133468.0	0.107618	0.809562	0.193031	0.0	0.0	0.0	0.000000

	year	month	count
0	2015	7	37
1	2015	8	113
2	2015	9	96
3	2015	10	104
4	2015	11	108
5	2015	12	91
6	2016	1	88
7	2016	2	90
8	2016	3	82
9	2016	4	88
10	2016	5	112
11	2016	6	86
12	2016	7	124
13	2016	8	205
14	2016	9	305
15	2016	10	164
16	2016	11	86
17	2016	12	87
18	2017	1	133
19	2017	2	241
20	2017	3	333
21	2017	4	297
22	2017	5	283
23	2017	6	266
24	2017	7	204
25	2017	8	240
26	2017	9	229
27	2017	10	188
28	2017	11	187
29	2017	12	138
30	2018	1	143
31	2018	2	143
32	2018	3	141
33	2018	4	147
34	2018	5	127
35	2018	6	186
36	2018	7	242
37	2018	8	126