Introduction

This project involves using Exploratory Data Analysis (EDA) to draw inferences on crime types and crime rates in Baltimore City, MD. Historical crimes data of Baltimore City, MD was obtained from Baltimore Open Data Website to explore the different type of crimes occurring in the city at different times of the year. It is important that the Baltimore police department are able to forecast crime types so that they will be able to allocate appropriate resources to the right neighborhood and district. The project identify the most important crimes that need special attention.

The Dataset

The Baltimore City crime data consist of all crime types that occur between October 1963 and March 2021. Data preprocessing performed involves removing all columns that has over 70% missing values, dropping unneeded columns, filling remaining missing values with the mode of each categorical variables.

	CrimeDateTime	Location	Description	District	Neighborhood	Latitude	Longitude	GeoLocation
0	2021/03/25 01:20:20+00	0 S CAREY ST	HOMICIDE	SOUTHERN	UNION SQUARE	39.2879	-76.6382	(39.2879,-76.6382)
1	2021/03/25 01:20:20+00	0 S CAREY ST	SHOOTING	SOUTHERN	UNION SQUARE	39.2879	-76.6382	(39.2879,-76.6382)
2	2021/03/24 00:08:00+00	4900 YORK RD	COMMON ASSAULT	NORTHERN	WINSTON-GOVANS	39.3480	-76.6096	(39.348,-76.6096)
3	2021/03/24 07:53:00+00	400 E PATAPSCO AVE	COMMON ASSAULT	SOUTHERN	BROOKLYN	39.2372	-76.6049	(39.2372,-76.6049)
4	2021/03/24 21:54:00+00	2500 GREENMOUNT AVE	ROBBERY - CARJACKING	NORTHERN	HARWOOD	39.3182	-76.6095	(39.3182,-76.6095)

From the raw dataset, the year 2021 contains the first 3 months while the data from 1963 to 2013 is very sparse with just few data recorded per year. Therefore, for this project, the dataset was subsetted from 2014 to 2020 which covers 7 years period.

bpd = bpd[(bpd['year']>=2014) & (bpd['year']!=2021)] 
bpd.groupby(['year','day_name']).size().unstack('day_name')

day_name	Friday	Monday	Saturday	Sunday	Thursday	Tuesday	Wednesday
year
2014	6826	6675	6334	6056	6359	6331	6487
2015	7029	7182	6582	6344	6906	7034	6757
2016	7289	7048	6610	6545	6899	7055	6984
2017	7747	7523	7234	7093	7310	7363	7550
2018	7167	7117	6551	6396	7064	7026	6857
2019	6700	6695	6530	6219	6432	6795	6616
2020	5414	5196	5299	5035	4952	4889	4961

Finally, the crime date was converted to pandas datetime and the index as datetimeindex. Some additional features were created such as year, month, day name and seasons.

Summary of preprocessing steps:

Checked for missing values and dropped all columns with missing values greater than 70%
Dropped columns (‘CrimeCode’,’Total_Incidents’,’Post’,’Inside_Outside’,’Premise’,’RowID’,’X’,’Y’) that are not necessary for this project
Dealt with the remaining missing values by replacing missing values in each column by the mode of the categorical variables and dropped the remaining missing values from the data
Converted crime date to datetime, split it into components such as year, month, time, day, hour, and minute.
Set the dataframe index to datetimeindex
Limit the data to the past 7 years covering 2014 to 2020 due to sparsity of the raw dataset
Extract the four meteorological seasons (spring, summer, fall and winter) from the crime month to show how crimes changes with the seasons

Exploratory Data Analysis (EDA)

Descriptive statistics

The following figures shows the relative proportion and distribution of all the crime types reported by the Baltimore PD.

treemap

crimes_dist

Among the 14 crime types reported, the top 5 crimes (Larceny-22.3%, Common assault-16.9%, Burglary-14.2%, Larceny from auto-12.9%, Agg. assault-11.3%) make up 77.6% of the total crimes

LARCENY                 22.273984
COMMON ASSAULT          16.865132
BURGLARY                14.164420
LARCENY FROM AUTO       12.935248
AGG. ASSAULT            11.263438
AUTO THEFT               8.714399
ROBBERY - STREET         6.902988
ROBBERY - COMMERCIAL     1.752909
SHOOTING                 1.398798
ROBBERY - RESIDENCE      1.058308
ROBBERY - CARJACKING     0.916539
HOMICIDE                 0.676958
RAPE                     0.652814
ARSON                    0.424066

all

From the figure above, you can see that there is a clear pattern in crime and the hour of the day, seasons and year. Generally, crime is low in the early morning hours from around 4:00 AM to 6:00 AM and it rises steadily to its peak in the late evening from 5:00PM. The crimes remains high throughout the night until it starts to decrease after midnight. It can also be seen that majority of the crime occur during summer when the weather is really nice for outdoor activities. The overall crime trend reveals that crimes start increasing gradually from 2014 until it reached the peak in 2017 and start decreasing steadily until reaching the lowest in 2020

districts

The northeast and southeast police district has the highest recorded overall crimes in Baltimore, while the western district has the lowest. The heatmap plot also reveals that fridays is the likely day that crimes occur in the city.

Safety within Baltimore Neighborhoods

This analysis was able to reveal that Downtown is the most dangerous neighborhood with crimes rate that was over 12,000 for the period under consideration. The safest neighborhoods all have crime rates less than 50 over the 7 years period. They include Taylor heights, Greenmount cemetery, Mt. pleasant park, Blythewood and Dundalk marine terminal.

neighborhood

Analyzing Individual Crime Type

Yearly trends of individual crime types

Examining the individual crime types reveals that only five crime types show increasing trends. Agg, assault increased conseiderably when comparing the count between 2014 and 2020. Homicide, Robbery-carjacking, shooting and Rape all display increased trends although they all make up a small fraction of the overall crime.

crime type trends

Summary of crime types with increasing trends

Crime Types	Increase from 2014 to 2020 (%)
Agg. assault	23
Homicide	70
Rape	8
Robbery-Carjacking	400
Shooting	100

For all the crime types with increasing trends, the following table shows the dangerous neighborhoods and police districts.

Crime type	Neighborhood	Police District
Agg. assault	Downtown	Northeast
Homicide	Sandtown-Winchester	Western
Rape	Downtown	Northeast
Robbery-carjacking	Frankford	Northwest
Shooting	Sandtown-Winchester	Western

Further analysis of the individual crimes with increasing trends reveals that summer is the most season when most of the crimes occurs (homicide, agg.assault, shooting) and spring/summer in the case of rape. For majority of the crime type (agg.assault, homicide, shooting and carjacking), the lowest occurrence is reported between 5am and 7am. It then starts to increase steadily until it reaches the peak between 9am and midnight. Rape on the other hand has the lowest occurrence between 6am and 11am with some few spikes around noon and 3pm. It then increases gradually until it reached the peak around midnight.

Conclusions

Analyzing 7 years (2014 to 2020) of Baltimore City crime data reveals that five crime types (Larceny, Common assault, Burglary, Larceny from auto, agg.assault) make up the majority of crime in the city, amounting to about 78% of the overall crime. Among the crime types reported, five (agg.assault, homicide, rape, robbery-carjacking and shooting) exhibit increasing trends with percentages increase that range from 8% to 400%. For most of these crime types, crime is low in the early morning hours from around 5am - 7am and rises steadily to its peak between 9pm and midnight. It also reveals that summer is the season when crime is likely to occur. Downtown is considered to be the most dangerous neighborhood because the crime count is highest. Rape on the other hand, usually occurs mostly around midnight and reduces gradually until it reaches the lowest in the early morning hours.

Check out full codes and notebook on github