Analyzing Crimes in Nigeria — Part 1
Looking at patterns in class of crimes committed, locations as well as people involved, we may be able to predict possible crimes events before they happen.
Reducing crimes in our nation, Nigeria, is of great importance. Tackling crimes with data-driven approach has proven to be successful with countries like the U.S. collaborating with Palantir using big data analysis to fight terrorism.
The objective of this analysis is look at the data we have, find patterns, make recommendations and possibly inform security stakeholders in Nigeria the need for data-driven policies for fighting crime.
Data collection does not have to be via data entries only, we could also work with existing crime records in prints using Optical Character Recognition technologies for data collection. Which will then be extracted and formatted for analysis
Points to Note
- In the last 3 years, crime rate in the country has risen by 68.31% according to Numbeo
- Crime and Safety indices in Nigeria are 71.38 and 28.62 respectively Numbeo
Now we deep dive into the analysis
- Years covered: 1995–2015
- Total number of crimes recorded: 8541 (This is may not be an accurate reflection)
- An average of 5 fatalies per crime and a standard deviation of 28 with the maximum reaching 1000
It is quite alarming to see the number of crimes committed rise interestingly over the years. What is more disturbing is exponential rise from 2011 to 2012.
This discovery leads us to exploring the rate of change on an annual basis.
Let’s extract months from our data sets and compare crimes committed month on month for the 1995 to 1999 and 2011 to 2015
Comparing death counts from the last two years of Military rule and years 2014 and 2015, it is very clear that we has lesser death counts viz-a-viz crime rates in the country during military regime. (You can also make further interpretations)
The top 15 locations from 2010 to 2015 where we have recorded the highest number of fatalities. It is very obvious that Maiduguri has recorded more 3500 deaths, followed by Baga with about 2000.
Let’s visualize our data on the map using geo points
Note: Areas with red markers indicate crime scene attributed to Boko Haram
Predicting the Next Crime Scene
After visualizing our data on map, let’s create a prediction model based on some clustering technique, KMeans Clustering. I’ll leave to reading about K Means Clustering as this will help your understanding of the algorithm.
As we do not want to bury ourselves in Curse of Dimensionality, we will only be working with few features in our dataset. Bear in the mind that this research is reproducable, hence you may consider playing with other features for better predictions.
- Crime actors (suspects)
- Type/class of crime (this will help know the right law enforcement agent to deploy)
- Latitude and Longitude (location)
- Day of the year
- Month of the year
Number of Clusters: 10
- The North-Western part of Nigeria recorded lesser crime rates in 2015. Hence, activities that encourage economic growths should be encouraged in these areas
- Security measures/policies implementated in the North-West can be copied and deployed to other geo-political zones
- The North-East and South-South have proved to be very peculiar. Further analysis can be done to better inform on proper policies to curtail crime in these areas.
- In addition, further analysis using Graph Modeling Techniques can also be done to uncover connections between crime scenes and suspects.
- Additional data points, such as population, unemployment rate, etc, may be considered to expand the scope of the analysis
There are lot’s of inferences to draw from this analysis, however I have only pointed out few.
Data Science and its applications for social good in Nigeria is still in its infancies while its applications towards our socio-economic transformation is massive and unlimited. Making our nation a safer place is collective effort as move towards our 2025 vision. #iPlayMyPartWithData
Learn more about DJI Innovation’s latest MAVIC PRO, DJI’s portable yet powerful personal drone by watching the video below:
When Kevin Stevens and I sat down to build out the 1H 2017 product roadmap at Choose Energy, we spent a fair amount of…therevoluzionne.com
2016 has been a very successful year for Artificial Intelligence and for Machine Learning, however both fields have an…therevoluzionne.com