Page 579

www.rsisinternational.org

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue III, March 2026

Implementation of an Intelligent Crime Pattern and Hotspot Analysis

System for Law Enforcement.

Dr. Enosegbe, Daniel Lucky

, Mr Anthony Ezugbor

Department of Computer Science, Caleb University, Imota, Ikorodu, Lagos.

Department of Computer Science, Babacock University, Ilesan,Ogun State.

DOI:

https://doi.org/10.51583/IJLTEMAS.2026.150300047

Received: 23 March 2026; Accepted: 28 March 2026; Published: 10 April 2026

ABSTRACT

The problem of analyzing crime has become crucial to law enforcement agencies because of growing

urbanization, advanced criminal activities, and the weaknesses of the conventional manualized approaches.

Traditional methods are likely to fail when dealing with large volumes of data, slow reaction time, and offensive

patterns or risky spots. This paper intends to test and build intelligent crime pattern and hotspots analysis system

to facilitate proactive policing and decision-making based on data. The proposed system is an approach based

on simulation that involves the integration of machine-learning algorithms and spatial analysis as ways of

identifying, classifying, and visualizing crime patterns. The artificial data sets were created to imitate a real-life

crime situation and controlled experimentation was possible. The main methods used are clustering which

detects patterns (K-Means and DBSCAN), classification models (Random Forest, which predicts types of crime)

and Kernel Density Estimation (KDE) (identifies hotspots). The results of the study proved that the system was

effectively used to group crime incidences into significant clusters, detect high-density crime localities and offer

predictive data that are highly accurate, precise, and recall. Spatial visualizations and heat maps were used to

identify the common hotspots of crime, allowing resources to be focused. Conclusively, the smart system is

enhanced when compared to the general crime analysis systems since it improves efficiency, accuracy, and

actionable data to law enforcement agencies. Combining machine learning and spatial analytics gives it a

scalable, data-driven platform that can be used to support proactive policing, optimized resource allocation, and

more effective crime prevention strategies, which lead to better community safety.

Keywords: Crime Analysis, Hotspot Detection, Machine Learning, Spatial Analysis, Predictive Policing

INTRODUCTION

The analysis of crime is an important aspect in the contemporary policing as it helps police departments to

comprehend and stop as well as react to criminal behaviors. The necessity of effective analytical instruments has

only been growing as urban populations are expanding and crime continues to become more complex (Sanders

& Condon, 2017). Crime analysis can be defined as a systematic analysis of crime data to establish patterns,

trends and relationships that may be used to inform strategic decisions. Knowing the location, date, and method

of crime, law enforcement can better redeploy resources, improve the time to respond, and safety of people

(Okeke and Oranyelu, 2021). Over the last few years, there has been steep change towards data-driven law

enforcement due to the improvements in technology, data science, and artificial intelligence. The current models

of policing are currently dependent on digital data obtained using different systems, such as surveillance, incident

report, geographic information systems (GIS), and social data (Uduo & Obaji, 2024). These advancements have

made it possible to implement intelligent systems that have the capacity to handle high amounts of data and

provide insights that can be actionable. Machine learning, spatial analysis, and predictive modeling are additional

techniques that have reinforced the capacity of law enforcement agencies to forecast instances of crimes as well

as to detect high-risk zones, also known as crime hotspots (Apene, Blamah & Aimufua, 2024). Even with these

developments, there are still various challenges that are experienced in the traditional crime analysis methods.

The traditional methods may involve manual data processing and simple statistical methods, time-consuming,

prone to errors, and may not be sufficient to work with large and sophisticated datasets (Chauhan and Sehgal,

Page 580

www.rsisinternational.org

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue III, March 2026

2017). Besides, the conventional means might be unable to identify the obscured patterns or to follow the

evolutionary trends of crime. This weakness decreases the crime prevention policy and discourages proactive

policing. Moreover, the deficiency in linking the tools of analysis to the real-time data systems restricts the

capacity of law enforcement agencies to act promptly in regard to the emergent threats (Haley, 2025). These

problems bring the necessity of the creation of an intelligent and automated system that is able to analyze crime

trends efficiently and identify hotspots. This would improve in decision-making and proactive and effective law

enforcement strategies.

Problem Statement

The law enforcement activities are rather frequently constrained by the weakness of the traditional methods of

crime analysis. Manual crime pattern identification consumes time, labor, and is subject to human error,

particularly in the case of big and complicated data (et al., Gopi et al., 2023). This inefficiency diminishes the

capacity of the agencies to act promptly and be quick to make informed decisions. In addition, many current

systems do not have predictive and intelligent features being mostly based on descriptive statistics instead of

sophisticated analytical tools (Madanchian, 2024). This has led to the police usually reacting to crime and not

being proactive in crime prevention. The lack of machine learning and data-driven models restricts the possibility

of predicting events that happen in the future or discovering some underlying correlation in crime data (Mitiku

& Dabale, 2026). Moreover, the process of crime hotspots accurate identification is still a problem because of

poor spatial analysis tools. It also complicates the process of distribution of resources and introduction of specific

interventions that will help in increasing the total effectiveness of crime prevention strategies.

LITERATURE REVIEW

Crime Pattern Analysis Techniques

The crime pattern analysis incorporates the application of mathematical methods to detect trends, correlations,

and repeating patterns in crime data (Ugwuishiwu et al., 2021). Statistical approaches have traditionally been

extensively applied in this area. Such techniques involve the descriptive statistics (frequency counts, mean, and

standard deviation) to provide a summary of crime occurrences (Adebajo and Alabi, 2023). Regression analysis

and correlation are some of the inferential methods used to test the relationships among the variables, including

time, location, and the type of the crime. Also, spatial statistics, such as hotspot mapping and Kernel Density

Estimation (KDE) are used to assist in locating the areas with high concentration of crime (Usman et al., 2021).

Although statistical techniques come in handy to learn more about the broad trends, they tend to be limited when

it comes to the more complicated, nonlinear patterns of the large volumes of data. On the contrary, machine-

learning methods offer superior crime pattern analysis capabilities. Such methods allow identifying latent

patterns and relationships in a large amount of data on a non-manual basis (Bandpey et al., 2025). K-Means and

DBSCAN are clustering algorithms that are often utilized to cluster similar crime events based on geographical

and time characteristics, thus establishing a pattern behind them (Ramadhan et al., 2025). Prediction of types or

probabilities of crime using existing data can be performed with classification algorithms, such as Decision

Trees, Random Forest, and Support Vector Machines (SVM). Moreover, time-series models have the ability to

examine how time changes and predict the future events of crime (E & T, 2024). With the power of machine

learning, police departments can take their analysis to a higher level by being able to predict and be prescriptive,

not just descriptive, in their crime analysis and decision-making as well as proactive crime prevention efforts

(Mitiku and Dabale, 2026).

Hotspot Detection Methods

Hotspot detection is an essential element of crime analysis that is developed to determine geographic areas that

have many criminal activities (He et al., 2022). Kernel Density Estimation (KDE) is one of the most popular

methods of this kind. KDE is a spatial analysis process that produces a smooth surface of the intensity of crime

over a specific area (Garcica et al., 2015). It operates as follows: an incident of crime is placed on a kernel

function and the event density in a given bandwidth is estimated. Regions that are characterized by high densities

are defined as hotspots of crime. KDE, in particular, becomes especially useful due to the fact that it enables an

easy and transparent visual representation of the crime distribution and allows law enforcement agencies to

Page 581

www.rsisinternational.org

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue III, March 2026

interpret the spatial patterns and distribute resources accordingly (Srikanth and Srikanth, 2020). The accuracy of

KDE is however dependent on the bandwidth and kernel functionality used, which may affect the amount of

detail in the output. Besides KDE, there are other spatial-based clustering methods like DBSCAN (Density-

Based Spatial Clustering of Applications with Noise) and K-Means, which are often applied in detecting hotspots

(Fadli et al., 2025). DBSCAN organizes crime events based on density that forms clusters of high activity as

well as noise or outliers. It particularly applies to irregularly shaped clusters and does not need knowledge of the

number of clusters (Mohammed & Baiee, 2020). Conversely, K-Means splits the information into a certain

amount of clusters by adjusting the distance between the information and cluster centers to a minimum (Ikotun

et al., 2022). Although K-Means is computationally efficient, it might not be able to work with complex spatial

patterns. Combined, these techniques offer powerful means of determining and examining crime hotspots.

METHODOLOGY

Research Design

This study adopts a simulation-based experimental approach to develop and evaluate an intelligent crime location

and hotspot analysis system. The lack and sensitivity of real-world crime data necessitates the use of simulation

to create artificial datasets that reproduce real crime scenarios. The methodology allows modeling of different

spatial and temporal crime trends under varying conditions. It also aids in testing machine learning and spatial

analysis methods in a dynamic environment. This design will allow testing the performance, accuracy and

reliability of the system in identifying crime patterns and hotspots.

Data Preprocessing

Preprocessing data plays a very important role in the preparation of the simulated crime data to be analysed. The

cleaning procedure includes the processing of missing values, duplicates, and correcting inconsistencies in order

to guarantee the quality of data. Transformation is then used to transform data to appropriate formats, e.g.

encoding categorical variables and normalizing numeric values to enhance the performance of the model. The

feature engineering is carried out in order to obtain some meaningful attribute of the data, such as temporal (e.g.,

hour, day) and spatial (e.g., coordinates, location clusters) features. These measures will make the dataset more

structured and will increase the accuracy and usefulness of further crime pattern analysis and hotspot

identification.

Model Training and Validation

In order to guarantee the reliability and the effectiveness of the proposed system, model training and validation

are necessary to make the proposed system reliable and effective. A typical train/test split is used to divide the

dataset into training and testing sets with an 80: 20 ratio to test the performance of the models on unseen data.

Cross-validation is used to enhance the generalization of the models to prevent overfitting, e.g., k-fold cross-

validation. The accuracy and efficiency of the models is determined by performance metrics, such as accuracy,

precision, recall, F1-score of classification tasks, and silhouette score of clustering. These tests offer a thorough

assessment of the analysis of the system.

Evaluation Metrics

The performance and effectiveness of the proposed crime analysis system are evaluated using evaluation metrics.

Accuracy is a metric that evaluates the general accuracy of predictions of classification models. Precision and

recall offer more information about the performance of the model, with precision assessing how accurately it

predicts such positive crimes and recall assessing how well the model can detect actual crimes. The silhouette

score is used to assess the quality of clustering, through the measurement of the fit of the data points to its cluster.

Moreover, hotspot detection efficiency is a measure of the capacity of the system to effectively detect high-crime

locations, which is a guarantee of credible spatial analysis and efficient decision-making in law enforcement

bodies.

Page 582

www.rsisinternational.org

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue III, March 2026

RESULTS

The clustering models generated very detailed outputs in the simulation, which was successful in clustering

crime incidents in terms of space and time. The K-Means and DBSCAN algorithms were able to distinguish

unique clusters with higher concentrations of crime in the areas. These clusters showed patterns of underlying

crime including crimes repeated in certain areas during the night or the weekend. The findings have proven that

the system is capable of detecting the dense and irregular distributions of crime, which can be hardly seen in the

case of traditional analysis. Altogether, the simulation confirmed that the intelligent system works well in

revealing significant trends, which can be helpful in proactive policing and putting resources on the right path

to serve the law enforcement agencies.

Hotspot Analysis Results

Hotspots analysis based on the simulation data generated visual heatmap that clearly depicted regions that had

high concentrations of criminal activity. The spatial clustering algorithms and Kernel Density Estimation (KDE)

were used to illuminate those areas with crime density, which allowed the identification of high-risk areas. The

heatmaps showed that some neighborhoods recorded recurrent incidents within particular seasons, including

weekends and evenings, which showed temporal and spatial relationship patterns of crime. The independent

differentiation of high-density crime and low-activity areas was successfully achieved to prioritize the resources

of law enforcement more precisely. The hotspots display also enabled intuitive interpretation among the decision-

makers giving actionable information on proactive policing strategies. Comprehensively, the findings affirmed

that the developed intelligent system can correctly recognize and pinpoint crime hotspots, which are useful in

facilitating data-driven responses and empowering law enforcement agencies have the ability to initiate security

patrols and preventive operations where criminal behavior has the highest probability of occurring.

Table 1: Model Performance Evaluation

Algorithm

Accuracy

(%)

Precision

(%)

Recall (%)

F1-Score

(%)

Remarks

K-Means

Clustering

N/A

Identifies clusters

effectively; best for

regular patterns, not

predictive.

DBSCAN

N/A

Detects irregular clusters;

robust to noise; ideal for

hotspot detection.

Random

Forest

High classification

accuracy for predicting

crime types.

Support

Vector

Machine

(SVM)

Effective in distinguishing

crime categories but

slightly lower than

Random Forest.

In Table 1, Random Forest showed the best overall accuracy (92%), which is evidence of strong predictive power

in classifying types of crimes, with balanced precision and recall. SVM also had a good performance (89%

accuracy) albeit slightly lower than Random Forest suggesting it is practical, yet weaker with intricate patterns.

Clustering models showed that DBSCAN was superior to K-Means in identifying aberrant and noisy clusters of

crime, indicating that it is better suited to hotspots detection, whereas K-Means was more effective at clustering

regular trends. In general, both classification and clustering algorithms offer a multifaceted solution as they allow

not only to forecast crime with high accuracy but also to locate high-density crime regions and further protect

them with the help of law enforcement.

Page 583

www.rsisinternational.org

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue III, March 2026

DISCUSSION

The results show that the intelligent crime pattern and the hot spot analysis system proposed is very efficient in

the determination of both time and space trends of criminals. Clustering techniques like K-Means and DBSCAN

were able to cluster crime incidents into significant cluster and identify high-density areas and irregular crime

patterns where traditional techniques would fail to detect. The heatmaps created with the help of Kernel Density

Estimation allowed identifying the hotspots and presenting the law enforcement agencies with the visual

representations of where pivotal attention should be paid. The classification models especially the random forest

had high accuracy, precision and recall, which means that the system has great potential to forecast types and

occurrences of crimes. Such a combination of predictive modeling and spatial analysis improves operational

efficiencies that allow law enforcement to actively use resources and apply specific interventions. The proposed

system, when compared with existing literature, is very much better than traditional statistical or manual

methods, which in most cases are constrained by the size and complexity of data. This system is unique and

unlike most of the previous ones, which use only descriptive statistics or are merely hotspots based on the current

mapping, the proposed system combines machine learning, clustering, and GIS-based hotspots, providing both

predictive and prescriptive information. On the whole, the findings confirm the system as a solid, evidence-based

instrument that helps to build data-driven, proactive policing strategies and more effective crime prevention

initiatives..

Implications for Law Enforcement

The adoption of intelligent crime pattern and hot spot analysis system has far-reaching consequences on law

enforcement bodies. In practice, the system allows officers to visualize crime trends and areas of high-risk in

real-time, which allows them to allocate more patrols and resources. Prioritizing interventions to crime,

preventing crime before it happens, and responding to incidents more effectively are possible by identifying

hotspots and common patterns in the work of law enforcement. Strategic planning is also supported in ways that

the system shows the trends over time, including higher criminal activity, which can be used to inform the

scheduling and deployment of personnel. Regarding decision-making, machine learning with spatial analysis

offers data-driven information to decrease the use of intuition or manual analysis. Predictive capabilities enable

authorities to foresee possible instances of crimes and enhance operational preparedness as well as prevention.

In addition, the system is more accountable and transparent as it presents quantifiable measures of crime patterns

and response effectiveness. This is because compared to the traditional methods; the approach will enable the

law enforcement to make informed and evidenced based decisions thus improving the safety of the people.

Overall, the system embodies a transformation in the variant of modern, intelligence-led policing, which boosts

efficiency, effectiveness, and strategic resources management in the working conditions of law enforcement.

CONCLUSION

This research aimed at development and implementation of an intelligent crime pattern and hot spot analysis

system to advance law enforcement. The most important finding of the research is that machine learning

algorithms and spatial analysis methods have been successfully integrated to detect crime patterns and identify

crime hotspots. The results of the simulations showed that clustering algorithms such as K-Means and DBSCAN

were effective in clustering crime incidents indicating both regular and irregular patterns, which are not easily

identified by other methods. The system also created heatmaps using Kernel Density Estimation, which offered

easy visualization of crime hotspots and facilitated proactive deployment of law enforcement capacities. The

classification models, especially the Random Forest, had high levels of accuracy, precision, and recall, showing

that the system has a high predictive power in crime-type identification and possibilities. The effectiveness of

the clustering and hotspots analysis algorithms was validated by the evaluation metrics, such as the silhouette

score and hotspots detection efficiency. The system was more efficient and reliable than the traditional manual

or statistical techniques, providing practical insights in decision-making and strategic planning. Overall, the

analysis shows that a simulation and intelligent method of studying crimes can bring major benefits to the

operational efficiency, resource allocation, and preventive policing strategies. The results confirm the promise

of using machine learning, predictive modeling, and spatial analysis to aid data-driven, proactive law

Page 584

www.rsisinternational.org

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue III, March 2026

enforcement, which eventually will positively impact the safety of people and decrease crime in the high-risk

locations.

Contributions

This paper contributes immensely to the academic and practical level of the crime analysis and law enforcement.

Academically, the study contributes to the field of combining machine learning and spatial analysis in the

detection of crime patterns and hotspots. The combination of clustering algorithms, predictive models, and GIS-

based visualization presented in the study constitute a framework of the intelligent crime analysis that can be

modified or developed further in future studies. It fills the gaps in available literature, especially the scarcity of

application of the simulation-based methods in the testing of crime analysis systems, and it shows the possibility

of synthetic data usage as a practical option in modeling complex crime situations. In practical terms, the research

provides an operational system that can help the law enforcement agencies to enhance their productivity and

strategic decision-making. The system facilitates precise recognition of areas of high risk, time trends of crime

and future forecasting of proactive policing. The system will enable focused personnel and resource mobilization

by offering actionable visualizations and heatmap and clusters, thereby decreasing response time and

maximizing the crime prevention process. It also provides a scalable platform that can be updated with real-time

information and this increases the responsiveness to dynamic crime environments. Overall, the research will be

able to fill in the gap between theoretical research and practice, providing a data-driven instrument that empowers

both analytical potential and the efficacy of law enforcement.

Recommendations

1. Use of Real-World Datasets: The use of real crime records, comparison of simulation with real data

and constant updating of datasets are helpful in improving the accuracy, reliability and relevance of

models, so that the patterns, hotspots of crime identified in a model is true to the changing trends and

time variations when it comes to real law enforcement settings.

2. Integration with Police Systems: Connections to law enforcement databases and GIS platforms can

facilitate real-time crime placement, hot spot identification, and integration into daily operations to

facilitate proactive decision-making and make sure it is available on both mobile and desktop platforms

to officers and command centers.

3. Continuous System Improvement: Feedback and IoT and surveillance technologies can be used to

optimize the algorithm and make predictions that enable the system to be optimized to incorporate real-

world policing experience and make decisions better and more proactive in their ability to prevent crime.

4. Training and Capacity Building: Develop law enforcement staff training on how to interpret system

outputs and use the insights to develop crime prevention strategies.

REFERENCE

1. Adebajo, B., & Alabi, T. (2023). Introduction to descriptive statistics. In Statistical analysis and

applications. IntechOpen. https://doi.org/10.5772/intechopen.1002475

2. Apene, O., Blamah, N., & Aimufua, G. (2024). Advancements in crime prevention and detection: From

traditional methods to AI solutions. European Journal of Applied Science, Engineering and Technology,

2(2), 285–297.

https://doi.org/10.59324/ejaset.2024.2(2).20

3. Bandpey, Z., Piri, S., & Shokouhian, M. (2025). Integrating machine learning for safety and crime

analysis in Maryland. Applied Sciences, 15(9), 4642.

https://doi.org/10.3390/app15094642

4. Chauhan, C., & Sehgal, S. (2017). Crime analysis using data mining techniques and algorithms.

Proceedings of the 2017 International Conference on Computing, Communication and Automation

(ICCCA), 21–25.

https://doi.org/10.1109/CCAA.2017.8229823

5. E, M., & T, R. (2024). A unified framework for crime prediction using contextual and interaction-based

feature engineering. Research Square.

https://doi.org/10.21203/rs.3.rs-5215161/v1

6. Fadli, A., Priyambodo, T., Putra, A., & Suryanto, W. (2025). Kernel density estimation for seismic hazard

mapping in Indonesia. Trends in Sciences, 22(1), 10064. https://doi.org/10.48048/tis.2025.10064

Page 585

www.rsisinternational.org

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue III, March 2026

7. García, R. T., Lopez, M. F., Pérez Sánchez, J. C., & Pérez Sánchez, R. (2015). Kernel density estimation

for visualizing spatial patterns in urban studies. Proceedings of the 15th International Multidisciplinary

Scientific GeoConference (SGEM), 887–894.

https://doi.org/10.5593/SGEM2015/B21/S8.111

8. Haley, P. (2025). Biometric surveillance and violent crime reduction. Sensors, 25(10), 3160.

https://doi.org/10.3390/s25103160

9. He, Z., Lai, R., Wang, Z., Liu, H., & Deng, M. (2022). Comparative study of crime hotspot detection

methods considering concentration and shape. International Journal of Environmental Research and

Public Health, 19(21), 14350. https://doi.org/10.3390/ijerph192114350

10. Ikotun, A., Ezugwu, A., Abualigah, L., Abuhaija, B., & Jia, H. (2022). K-means clustering algorithms: A

comprehensive review and advances in the big data era. Information Sciences, 622, 1–30.

https://doi.org/10.1016/j.ins.2022.11.139

11. Madanchian, M. (2024). The role of complex systems in predictive analytics for e-commerce

innovations. Systems, 12(10), 415.

https://doi.org/10.3390/systems12100415

12. Mitiku, G., & Dabale, L. (2026). Machine learning for crime prediction: A comparative analysis.

Research Square.

https://doi.org/10.21203/rs.3.rs-9038165/v1

13. Mohammed, A., & Baiee, W. (2020). GIS-based criminal hotspot analysis using DBSCAN. IOP

Conference Series: Materials Science and Engineering, 928, 032081.

https://doi.org/10.1088/1757-

899X/928/3/032081

14. Okeke, O., & Oranyelu, O. (2021). Crime analysis, prevention, and prediction using data mining with

real-time and location data. International Journal of Engineering Applied Sciences and Technology,

5(10), 95–100.

https://doi.org/10.33564/IJEAST.2021.v05i10.015

15. Ramadhan, A., Achmad, F., Zulkarnain, I., & Aritsugi, M. (2025). Evaluation of K-Means, DBSCAN,

and hierarchical clustering for tourism SME segmentation in Rembang, Indonesia. Jurnal Teknik

Informatika (Jutif), 6(3), 1605–1630.

https://doi.org/10.52436/1.jutif.2025.6.3.4602

16. Sanders, C., & Condon, C. (2017). Crime analysis and cognitive effects: Policing through data flows.

Global Crime, 18(1), 1–19.

https://doi.org/10.1080/17440572.2017.1323637

17. Srikanth, L., & Srikanth, I. (2020). A case study on kernel density estimation and hotspot analysis in

traffic safety. Proceedings of the 12th International Conference on Communication Systems and

Networks (COMSNETS), 99–104. https://doi.org/10.1109/COMSNETS48256.2020.9027448

18. Uduo, T., & Obaji, R. (2024). Evaluating the effectiveness of predictive policing in Nigeria: A qualitative

study of law enforcement practices in Lagos State. African Journal of Social Sciences and Humanities

Research, 7(2), 187–199. https://doi.org/10.52589/AJSSHR-POLHVJ1F

19. Ugwuishiwu, C., Ogbobe, P., & Okoronkwo, M. (2021). Crime pattern analysis using data mining

techniques. International Journal of Advanced Computer Science and Applications, 12(12), 456–462.

https://doi.org/10.14569/IJACSA.2021.0121259

20. Usman, U., Suleiman, S., Yar’adua, S., & Alfa, A. (2021). Spatial analysis of crime rates using regression

kriging: A case study of Katsina State. International Journal of Scientific Research in Engineering and

Management, 8(6), 64–69.