Overview

This assignment is an analysis of crime data collated by the Houston Police Department in the U.S. state of Texas. The sample dataset used in this example was created by Prof. Chris Brunsdon of the National Centre for Geocomputation (NCG) at Maynooth University.

Graph of Total Crimes

The first step in analysing the sample data is to create an overview graph showing the number of crimes in thousands and divided by category.

From the overview graph it is clear that the categories of ‘Murder’ and ‘Rape’ are the smallest, while ‘Theft’ and ‘Burglary’ are the largest. From this initial graph, this would appear that the rates of violent and sexual crimes are low in the city of Houston, TX, with the exception of ‘Aggravated Assaults’.

However as this data is only a sample, this conclusion is tentative and inconclusive at present. To non-U.S. citizens the different variants in the types of crimes outlined in this dataset may seem confusing as seemingly different crimes are categorised in unique ways depending on the jurisdiction. For example, ‘Burglary’, ‘Theft’, ‘Robbery’ and ‘Auto-Theft’ all involve taking something from someone, however each category refers to a distinct crime. In addition, due to compound sentencing practices within the U.S. justice system, a criminal may be charged with combinations of different crimes. This has an impact on the data reported, as some criminals may have been charged with ‘Robbery’, ‘Theft’ and ‘Auto-Theft’ for the same criminal act. This adds to an artificial inflation of crime statistics for certain crimes.

For the purposes of this analysis, the crimes data set was subdivided into three different categories.

    Murders
    Theft
    Aggravated Assault

These specific crimes were chosen to provide an average, as thefts are the single largest crime sample and murder are the smallest, with aggravated assaults being the only non-theft category that is between thefts and murders in terms of number of crimes committed.

In addition two types of graphs were created + Days of the Week + Time of Day

Murders by Days of the Week

From the above graph, it would appear that murders spike over the weekend and are lowest in the middle of the working week. However these results are based upon police records, which means that the crimes are recorded when the officer or police station are informed about the crime. Therefore, certain crimes such as Theft cannot be fully viewed as representative when graphed by day of the week. For example, an individual’s home may have been broken into over the weekend but not discovered until the following week.

Thefts by Days of the Week

This graph shows that thefts are evenly distributed throughout the week with a slightly drop on Sundays. However as stated earlier, these results cannot be viewed as definitive.

Aggravated Assaults by Days of the Week

The Aggravated Assault graph shows that weekends have a higher rate of assaults than mid-week. This is indicative of higher rates of violence occurring over weekends, which is the traditional time off work for the majority of the populace. Therefore, assaults could be described as spiking during social time periods. This can be better explored via time specific graphs.

Conclusion

From the graphs outlining crimes occurring by days of the week, it is possible to draw tentative conclusions from the crimes data. However a more discursive approach is to examine crimes by time of day, this provides a more coherent overview of crimes occurring and could result in likely times for when certain crimes occur. The concept of time specific crimes is a complex one as it is often difficult to ascertain the specific time a criminal act was committed, but it has been used in a series of public safety campaigns particularly in the United States over the past few decades.

Murders by Time of Day

From the graph of time of day two conclusions are available + Murders are more likely to occur between 9pm - 3am + Murders are least likely to occur between 7am - 10am

These findings while interesting are conditional, it is unclear if the spike of murders at 11am is indicative of police being called to the site of a murder that occurred earlier. When examining murder rates, a more appropriate measure would utilise a mixture of police data which records when an officer learned of the crime, and coroner’s reports which often can estimate a time of death. The murders data is more appropriately explored via interactive maps, as this allows for a more granular overview of where murders are occurring, and by comparing racial and ethnic breakdowns of the city of Houston, TX.

Thefts by Time of Day

This graph shows thefts by time of day. From this graph the most interesting finding is that thefts are least likely to occur between 1am - 7 am. There are peaks and valleys throughout the day after this, with the most likely time for thefts to occur between 5pm - 7pm. In a similar manner to murder rates, this information is conditional upon when the crime was reported and the time that officers responded to the crime. This would partly explain the unusually low crime period between 1am - 7am, as there would be a delayed discovery and reporting of the crime at that time of day.

Aggravated Assaults by Time of Day

The graph of Aggravated assaults shows that the most likely time for these crimes to occur is between 9pm - midnight. This corresponds with the finding that assaults are more likely to occur over the weekend and during the hours when people would be socialising; going to bars, restaurants and other social gathering spaces.

Conclusion

The use of graphs to analyse thefts and aggravated assaults illustrate different attributes of both crime types and allows for tentative conclusions about the time periods when these crimes are most likely to occur. In the next section, the focus will be upon murders and will focus upon the geographical dispersement of murders throughout the city of Houston.

Mapping Murders

Using R, it is possible to add longitude and latitude data to the addresses of crime categories. However due to the rules surrounding the use of a free Open Street Maps service, it was only possible to create interactive leaflet maps using the murder category.

As shown in the map below, murders can be portrayed in a generalised way without distinction. However without colour coding it is a limited measure from the perspective of analysis.

Therefore the murders category was divided into subcategories of days of the week.

Monday Murders in Houston

This map shows the murder rate in Houston on Mondays and the time. This map combines much of the findings of the earlier day of the week and time of day graphs and presents the data in a easily viewable interactive map. The colour coding of a single day allows for greater clarity.

Monday - Friday Murders in Houston

This map shows the murder rate in Houston during the week. Each day is colour coded differently and can be selected using the sidebar created. Therefore a map of Monday and Thursday murders can be viewed each day individually, or as the entire week. The most interesting finding from this map is that the geographical dispersement is largely similar in the city of Houston, regardless of the day. The clusters of murders colour coded by different days are indicators that specific neighbourhoods within Houston have far higher murder rates, with multiple murders occurring within a given week.

Weekend Murders in Houston

This map shows the murder rate in Houston over the weekend. This map was created to explore the finding that the murder rate in Houston is higher over the weekend. However, the geographic dispersement of the murders across the city differs slightly depending on what days are being compared. For example, comparing Monday murders to Saturday murders highlights different neighbourhoods in different parts of the city. The clustering effect is much more pronounced in this map due to the earlier finding that murders are higher on the weekend.

Total Murders in Houston

This final map is of murders in Houston, TX coloured coded for weekdays and weekends. This allow for a direct day to day comparison geographically across the city of Houston.

Murders by Time of Day

This last map is an example of using time to pinpoint when crimes, in this case, murders occurred. Although it is interesting and visually striking, the results are useful only as a guide.

Summation

In conclusion, the crime data from Houston can be broken down into categories and examined individually. Using a mixed approach of graphs and leaflet maps provides for more accurate findings from the data. From this brief analysis, it is clear that the usage of crime data in conjunction with demographic, economic and social data about the city of Houston could lead to interesting findings about the relationship between the time and days when crimes occur and the neighbours which have corresponding high rates of specific crimes.