Using FB'sProphet to forecast crime rate in Chicago using historical data from 2005-2017

chicago_crimeforecast.py

# -*- coding: utf-8 -*-
"""
Created on Wed Jun  3 21:13:54 2020

@author: Localuser
"""

import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
from fbprophet import Prophet

chicago_df_1 = pd.read_csv('Chicago_Crimes_2005_to_2007.csv', error_bad_lines = False)
chicago_df_2 = pd.read_csv('Chicago_Crimes_2008_to_2011.csv', error_bad_lines = False)
chicago_df_3 = pd.read_csv('Chicago_Crimes_2012_to_2017.csv', error_bad_lines = False)

chicago_df = pd.concat([chicago_df_1,chicago_df_2, chicago_df_3])

plt.figure(figsize = (10,10))
sns.heatmap(chicago_df.isnull(), cbar = False, cmap = 'YlGnBu')

chicago_df.drop(['Unnamed: 0', 'Case Number', 'ID', 'IUCR', 'X Coordinate', 'Y Coordinate', 'Updated On', 'Year', 'FBI Code', 'Beat', 'Ward', 'Community Area', 'Location', 'District', 'Latitude', 'Longitude'], inplace = True, axis = 1)

chicago_df.Date = pd.to_datetime(chicago_df.Date, format = '%m/%d/%Y %I:%M:%S: %p')
chicago_df.index = pd.DateTimeIndex(chicago_df.Date)

order_data = chicago_df['Primary Type'].value_counts().iloc[:15].index

plt.figure(figsize = (15, 10))
sns.countplot(y = 'Primary Type', data = chicago_df, order = order_data)

plt.figure(figsize = (15, 10))
sns.countplot(y = 'Location Description', data = chicago_df, order = chicago_df['Location Description'].value_counts().iloc[:15].index)

chicago_df.resample('Y').size()

plt.plot(chicago_df.resample('Y').size())
plt.title('Crime count per year')
plt.xlabel('Years')
plt.ylabel('Number of Crimes')

plt.plot(chicago_df.resample('M').size())
plt.title('Crime count per month')
plt.xlabel('Months')
plt.ylabel('Number of Crimes')

chicago_prophet = chicago_df.resample('M').size().reset_index()

chicago_prophet.columns = ['Date', 'Crime Count']
chicago_prophet_df_final = chicago_prophet.rename(columns = {'Date':'ds', 'Crime Count': 'y'})

m = Prophet()
m.fit(chicago_prophet_df_final)

future = m.make_future_dataframe(periods = 365)
forecast = m.predict(future)

figure = m.plot(forecast, xlabel = 'Date', ylabel = 'Crime Rate')
figure = m.plot_components(forecast)