THE UNIVERSITY OF ZAMBIA

IN CONJUNCTION WITH
ZAMBIA UNIVERSITY
COLLEGE OF TECHNOLOGY

PROJECT PROPOSAL

Topic
Developing a Crime Prediction System using Machine Learning and Historical Crime Data

A case of Ndola Police Department

Tembo Annie P - 1910585

Bulaya Kapasa Katherine - 2300667
Naonga kaluba - 2300694
Kalala Getrude - 2300685

@2025
Table of Content
 Introduction

Overview
This project aims to help the law enforcement agencies to identify potential crime hotspots and
take proactive measures to prevent the respective crimes. The study aims to use machine learning
algorithms to analyze historical crime data and predict future crime patterns. The system will
collect historical crime data, including:
1. Crime Type: Type of crime (e.g., theft, assault, robbery)
2. Location: Location of crime (e.g., address, latitude, longitude)
3. Time: Time of crime (e.g., date, time of day)
4. demographics (Age distribution, ethnic diversity, population density, income levels)

Background
The increasing complexity and dynamic nature of crime patterns pose significant challenges for
law enforcement agencies worldwide. Traditional crime prevention strategies often rely on
reactive approaches, which can be ineffective in addressing the root causes of crime. This project
aims to develop a proactive crime prediction system that leverages machine learning algorithms to
analyze historical crime data and predict future crime patterns.
Crime prediction and prevention have become critical concerns for law enforcement agencies, with
the goal of reducing crime rates and improving public safety. Recent studies have demonstrated
the potential of machine learning algorithms in crime prediction, including the use of spatial-
temporal analysis and demographic data [1], [2].
The integration of machine learning techniques with crime data analysis has shown promising
results in identifying crime patterns, hotspots, and trends, enabling law enforcement agencies to
take proactive measures to prevent crimes.
The application of machine learning in crime prediction involves various techniques, including
supervised and unsupervised learning algorithms. Supervised learning algorithms, such as
regression and classification models, can be used to predict crime rates and identify high-crime
areas [3]. Unsupervised learning algorithms, such as clustering models, can be used to identify
patterns and anomalies in crime data [4]. Additionally, deep learning techniques, such as neural
networks, have been used to analyze complex crime data and predict crime patterns [5].
The use of machine learning in crime prediction has several benefits, including improved accuracy,
efficiency, and effectiveness. Machine learning algorithms can analyze large datasets and identify
complex patterns that may not be apparent through traditional analysis. This enables law
enforcement agencies to allocate resources more effectively and take proactive measures to
prevent crimes.
Machine learning plays a significant role in crime prediction, enabling law enforcement agencies
to identify potential crime hotspots and take proactive measures. The integration of machine
learning techniques with crime data analysis has shown promising results in identifying crime
patterns, hotspots, and trends [1]. Supervised learning algorithms, such as logistic regression and
decision trees, can be used to predict crime rates and identify high-crime areas [2]. Unsupervised
learning algorithms, such as clustering models, can be used to identify patterns and anomalies in
crime data [3].
Deep learning techniques, such as convolutional neural networks (CNNs) and recurrent neural
networks (RNNs), have been used to analyze complex crime data and predict crime patterns [4].
CNNs can analyze spatial patterns in crime data, while RNNs can analyze temporal patterns. A
deep learning-based multi-modal surveillance system can integrate multiple data sources, such as
crime data, demographic data, and weather data, to predict crime patterns [5].
The use of machine learning in crime prediction has several benefits, including improved accuracy,
efficiency, and effectiveness. Machine learning algorithms can analyze large datasets and identify
complex patterns that may not be apparent through traditional analysis [6]. However, there are also
challenges associated with using machine learning in crime prediction, such as data quality issues
and interpretability [7]. Machine learning models require high-quality data to produce accurate
results, and complex models can be difficult to interpret, making it challenging to understand the
reasoning behind predictions.
The proposed system has the potential to significantly improve crime prevention and public safety.
By predicting crime hotspots and patterns, law enforcement agencies can allocate resources more
effectively, reducing the likelihood of crimes occurring. This project builds on existing research
in crime prediction and machine learning, and contributes to the development of more effective
crime prevention strategies [3].
Problem statement
The Ndola police department is confronted with significant challenges that hinder its ability to
effectively predict and prevent crimes, ultimately compromising public safety. One of the primary
concerns is the inefficiency of current crime prediction and prevention methods, which has led to
increased crime rates. The manual analysis of crime data is a time-consuming and labor-intensive
process that often yields inaccurate results, making it difficult for law enforcement agencies to
identify crime patterns and hotspots.
Furthermore, the lack of data-driven insights hinders the ability of law enforcement agencies to
make informed decisions and optimize resource allocation. Without access to accurate and timely
data, police departments are unable to effectively deploy resources, respond to emerging trends,
and prevent crimes.

Research Aim
The aim of this research is to develop a crime prediction system using machine learning algorithms
that can accurately forecast crime trends and patterns in Ndola, Zambia, and provide law
enforcement agencies with a proactive tool to predict and prevent crimes, ultimately enhancing
public safety.

Research Objectives
1. To enhance public safety by providing warning alerts of crime hotspots areas and trends to
the public
2. To provide a platform for law enforcement agencies that helps with data-driven insights to
inform decision-making.
3. To develop a system that implements a machine learning model that can accurately predict
crime hotspots and types.
Research Questions
1. Can machine learning algorithms enhance public safety?
2. How can the crime prediction system be integrated with existing law enforcement systems
to inform decision-making?
3. Can machine learning algorithms accurately predict crime hotspots and types using
historical crime data?

Scope of Study
This study focuses on developing a crime prediction system for the Ndola police department using
machine learning algorithms and data analysis techniques. The scope of the study includes the
collection and analysis of historical crime data, identification of crime patterns and hotspots, and
development of a predictive model that can forecast future crime trends.
The study will concentrate on crimes reported in Ndola, Zambia, and will utilize data from various
sources, including law enforcement agencies and government databases. The predictive model will
be developed using machine learning algorithms, such as Random Forest and Support Vector
Machines, and will be evaluated using metrics such as accuracy, precision, and recall.
The scope of this study also includes the development of a user-friendly interface that can be used
by law enforcement agencies to visualize crime patterns and trends, and to make informed
decisions about resource allocation and crime prevention strategies. The study aims to provide a
comprehensive solution for crime prediction and prevention in Ndola, and to contribute to the
existing body of research on crime prediction using machine learning techniques.
Significant of study
This study has significant implications for the Ndola police department and the broader
community. By developing a crime prediction system using machine learning algorithms, this
study aims to provide law enforcement agencies with a proactive tool to predict and prevent crimes,
ultimately enhancing public safety.
The significance of this study lies in its potential to improve the efficiency and effectiveness of
crime prevention strategies. By identifying crime patterns and hotspots, law enforcement agencies
can optimize resource allocation, respond quickly to emerging trends, and prevent crimes before
they occur.
Furthermore, this study contributes to the existing body of research on crime prediction using
machine learning techniques, providing valuable insights into the application of these techniques
in the context of Ndola, Zambia. The findings of this study can inform policy decisions and
resource allocation, ultimately leading to a safer and more secure community.
The study's significance also extends to the broader community, as it has the potential to improve
the quality of life for residents of Ndola. By reducing crime rates and improving public safety, this
study can contribute to economic growth, social stability, and overall well-being of the community.
 Literature Review
Overview
The terms “artificial intelligence” and “machine learning” are often incorrectly used
interchangeably [23]. Simply put, machine learning is a subset of artificial intelligence, in which
algorithms learn from input data to make predictions and identify patterns. With this seemingly
vague definition, it is often beneficial to break up the algorithms into a spectrum, between either
fully human-guided or fully machine guided analyses [24]. The use of machine learning algorithms
in crime prediction and mapping has gained significant attention in recent years. According to
Ngoge et al. (2024), machine learning enables computers to automatically learn from experience
without being explicitly designed, making it a valuable tool for crime analysis [8]. The authors
developed a machine learning model that combined time, space, and contextual information to
improve crime prediction and mapping.

Machine learning algorithms can be broadly categorized into two major types: supervised and
unsupervised learning. Supervised learning algorithms are able to forecast classes using the
features that are present in the data [9]. According to Kanimozhi et al. (2021), multiclass target
variables can be classified using supervised prediction by comparing the accuracies of various
classification algorithms [10]. Unsupervised learning models, on the other hand, classify or cluster
inconsistent, unstructured datasets.
Fig. 1. Machine learning can be divided into supervised, unsupervised, and reinforcement learning.
These three sub-fields each have their own applications

Model Development and Evaluation

Assessment of model quality occurs informally throughout the model development process. For
instance, when constructing a model, the aim is not to produce just any model of the target system
but to produce a good model, and this informs the choices made. Model evaluation, however, is
also frequently identified as a distinct step in model development, occurring after a model has been
fully constructed. Crime prediction is a process where a model uses different algorithms to solve
classification problems based on historical data. Using machine learning, these models can predict
and classify target variables [11]. According to Llaha (2020), classification models are expected
to input an unseen dataset and correctly predict category labels [5]. Techniques such as Linear
Regression, Decision Trees, Random Forests, Naive Bayes, K-Nearest Neighbors, and Support
Vector Machines can be used for crime prediction. In the study conducted by [12], the crime data
used to develop the model was collected from various sources, including law enforcement
organizations and websites [12]. The data was preprocessed into a suitable form to improve the
classification of rows/instances. The random forest algorithm emerged as the best algorithm with
a classification accuracy of 97% or 0.973301 [12]. The model was able to predict crime categories
and visualize their occurrence locations using contextual features. Visualization of crime was done
and presented using interactive plots such as bar graphs, line graphs, pie charts, and maps [12]. By
preprocessing this data into a suitable form, law enforcement agencies can use machine learning
models to analyze and depict crime occurrence locations.
Fig. 1. A typical workflow for machine learning model creation, evaluation, and deployment.
Once prepared, data are typically split into training, validation, and test sets. Training data are
typically used to create the model and choose the algorithm that performs best, whereas validation
sets are used for hyperparameter selection for model refinement. The model should then be
evaluated on test data sets, which have been blinded to model creation before deployment into a
usable model for predictions. The model should lastly be monitored and retrained for maintenance
with the option of deployment to another site for external validation

Linear Regression
A linear model makes a prediction by simply computing a weighted sum of the input features, plus
a constant called the bias term (also called the intercept term), as shown in Equation below.
Linear Regression model prediction
ŷ = θ0 + θ1 X1 + θ2 X2 + ⋯ + θn Xn
ŷ is the predicted value.
n is the number of features.
xi is the ith feature value.
θj is the jth model parameter (including the bias term θ0 and the feature weights θ1, θ2, ⋯, θn).

The next step is to select a performance measure. A typical performance measure for regression
problems is the Root Mean Square Error (RMSE). It gives an idea of how much error the system
typically makes in its predictions, with a higher weight for large errors. Equation below shows the
mathematical formula to compute the RMSE.
This equation introduces several very common Machine Learning notations that we will use
throughout this study:
1. m is the number of instances in the dataset you are measuring the RMSE on.
For example, if you are evaluating the RMSE on a validation set of 20 neighborhoods
or towns in a district, then m = 20.
2. x is a vector of all the feature values (excluding the label) of the ith instance in the dataset,
(i)

and y(i) is its label (the desired output value for that instance).
For example, if the first neighborhoods or towns in the dataset is located at longitude –
118.29°, latitude 33.91°, and it has 1,416 inhabitants with time of crimes at nighttime
(10PM - 02AM) and the number of crimes committed in a specific timeframe, that is,
last month is 156 (ignoring the other features for now), then:

−118.29
33.91
x(1) = ( )
1,416
10 − 02

And

y(1) = 156

3. X is a matrix containing all the feature values (excluding labels) of all instances in the
dataset. There is one row per instance and the ith row is equal to the transpose of x(i), noted
(x(i))T. For example, if the first neighborhood is as just described, then the matrix X looks
like this

(x(1))T
(x(2))T
−118.29 33.91 1,416 10 − 02
X= ⋮ =( )
(x(19))T ⋮ ⋮ ⋮ ⋮
((x(20))T)

Recall that the transpose operator flips a column vector into a row vector (and vice versa).
 4. h is your system’s prediction function, also called a hypothesis. When the system is given
an instance’s feature vector x(i), it outputs a predicted value ŷ(i) = h(x(i)) for that instance
(ŷ is pronounced “y-hat”).
For example, if your system predicts that the number of crimes committed in a specific
timeframe, that is, last month is 158, then ŷ(1) = h(x(1)) = 158. The prediction error for
this district is ŷ(1) – y(1) = 158 – 156 = 2.
RMSE(X,h) is the cost function measured on the set of examples using the hypothesis
h

Decision Trees
Decision Trees are versatile Machine Learning algorithms that can perform both classification and
regression tasks, and even multioutput tasks. They are very powerful algorithms, capable of fitting
complex datasets. Decision Trees are fairly intuitive and their decisions are easy to interpret. Such
models are often called white box models. In contrast, Random Forests or neural networks are
generally considered black box models. They make great predictions, and you can easily check the
calculations that they performed to make these predictions; nevertheless, it is usually hard to
explain in simple terms why the predictions were made. For example, if a neural network says that
a particular person appears on a picture, it is hard to know what actually contributed to this
prediction: did the model recognize that person’s eyes? Her mouth? Her nose? Her shoes? Or even
the couch that she was sitting on? Conversely, Decision Trees provide nice and simple
classification rules that can even be applied manually if need be (e.g., for Criminal classification).
A Decision Tree can also estimate the probability that an instance belongs to a particular class k:
first it traverses the tree to find the leaf node for this instance, and then it returns the ratio of training
instances of class k in this node.

Ensemble Learning and Random Forests

If you aggregate the predictions of a group of predictors (such as classifiers or regressors), you
will often get better predictions than with the best individual predictor. A group of predictors is
called an ensemble; thus, this technique is called Ensemble Learning, and an Ensemble Learning
algorithm is called an Ensemble method. For example, you can train a group of Decision Tree
classifiers, each on a different random subset of the training set. To make predictions, you just
obtain the predictions of all individual trees, then predict the class that gets the most votes. Such
an ensemble of Decision Trees is called a Random Forest, and despite its simplicity, this is one of
the most powerful Machine Learning algorithms available today. Moreover, often use Ensemble
methods near the end of a project, once you have already built a few good predictors, to combine
them into an even better predictor
Suppose you have trained a few classifiers, each one achieving about 80% accuracy. You may
have a Logistic Regression classifier, an SVM classifier, a Random Forest classifier, a K-Nearest
Neighbors classifier, and perhaps a few more
Fig 3. Training diverse classifiers
A very simple way to create an even better classifier is to aggregate the predictions of each
classifier and predict the class that gets the most votes. This majority-vote classifier is called a
hard voting classifier (see the figure below)

Fig 4. Hard voting classifier predictions

A Random Forest is an ensemble of Decision Trees, generally trained via the bagging method (or
sometimes pasting), typically with max_samples set to the size of the training set. The Random
Forest algorithm introduces extra randomness when growing trees; instead of searching for the
very best feature when splitting a node, it searches for the best feature among a random subset of
features. This results in a greater tree diversity, which (once again) trades a higher bias for a lower
variance, generally yielding an overall better model.
Related Studies
S. Venkatesh et al, 2024: Crime Prediction Using Machine Learning and Deep Learning
Crime prediction has emerged as a critical application of machine learning (ML) and deep learning
(DL) techniques, aimed at assisting law enforcement agencies in reducing criminal activities and
improving public safety. This case study focuses on developing a robust crime prediction system
that leverages the power of both ML and DL algorithms to analyze historical crime data and predict
potential future incidents.
The proposed system integrates a combination of classification and clustering techniques to
identify crime-prone areas, trends, and patterns. Machine learning algorithms like Decision Trees,
Random Forest, and Support Vector Machines are used for efficient data preprocessing and pattern
recognition. Deep learning models, including Convolutional Neural Networks (CNN) and
Recurrent Neural Networks (RNN), enable advanced feature extraction and temporal analysis [13].
The system's performance is evaluated using metrics such as precision, recall, F1-score, and
accuracy, demonstrating its reliability and scalability for real-world applications. The use of
visualization tools allows stakeholders to comprehend crime patterns effectively, supporting
proactive policing strategies.
While the proposed system shows promising results, there are some gaps that need to be addressed:

1. Data Quality: The accuracy of the system depends on the quality of the historical crime data.
Incomplete or biased data can lead to inaccurate predictions.
2. Feature Engineering: The selection of relevant features and parameters, such as time, location,
type of crime, and socio-economic factors, is crucial for the system's performance.
3. Interpretability: The use of complex machine learning and deep learning models can make it
challenging to interpret the results and understand the underlying patterns.
The proposed crime prediction system has the potential to revolutionize crime management by
transforming traditional reactive measures into predictive, preventive, and data-driven approaches.
Further research can focus on addressing the gaps identified above and exploring new techniques
to improve the system's performance and applicability.

S. Sridharan et al, 2024: Crime Prediction using Machine Learning

Crime prediction is a critical application of machine learning that aims to identify crime patterns
and trends to find underlying issues and potential solutions to crime. This case study focuses on
predicting crime cases from 2017 to 2020 using a dataset from 2001 to 2016.
The study uses machine learning algorithms such as Linear Regression and Random Forest
Classifier to predict crime trends [14]. The dataset is analyzed to identify trend-changing years,
and projections are made for each state as well as all states in India.
The study demonstrates the effectiveness of machine learning in crime prediction and highlights
the importance of data visualization in representing crime trends. Simple visualization charts are
used to represent the projections, making it easier to understand the crime patterns.
The study shows that machine learning can be a valuable tool in crime prediction and prevention.
By analyzing historical crime data, law enforcement authorities can more efficiently allocate
resources and target initiatives to reduce crime and increase public safety.
While the study demonstrates the potential of machine learning in crime prediction, there are some
gaps that need to be addressed:

1. Data Quality: The accuracy of the predictions depends on the quality of the historical crime
data.
2. Feature Engineering: The selection of relevant features and parameters is crucial for the
performance of the machine learning models.
3. Interpretability: The use of complex machine learning models can make it challenging to
interpret the results and understand the underlying patterns.

The use of machine learning algorithms in crime prediction and mapping has shown promising
results. By combining time, space, and contextual information, machine learning models can
predict crime categories and visualize their occurrence locations. Further research can be done to
improve the accuracy and performance of these models.

Theoretical framework
The theoretical framework for this study is based on the following theories and concepts:
1. Crime Pattern Theory
This theory suggests that crimes are not randomly distributed, but rather follow patterns and trends
that can be identified and analyzed [15]. The theory provides a foundation for understanding the
spatial and temporal distribution of crimes.
2. Routine Activity Theory
This theory posits that crime occurs when there is a convergence of motivated offenders, suitable
targets, and a lack of capable guardianship [16]. The theory highlights the importance of
understanding the environmental and social factors that contribute to crime.
3. Machine Learning Theory
Machine learning theory provides a framework for developing predictive models that can learn
from data and make accurate predictions [17]. The theory underlies the development of the crime
prediction system in this study.
Conceptual Framework
The conceptual framework for this study consists of the following components:
Input Variables
Historical crime data (crime type, location, time, etc.)
Socio-economic factors (poverty rate, unemployment rate, etc.)
Demographic factors (population density, age distribution, etc.)

Machine Learning Model

Data preprocessing (data cleaning, feature scaling, etc.)
Feature selection (identifying relevant variables)
Model training (Random Forest, Support Vector Machines, etc.)
Model evaluation (accuracy, precision, recall, etc.)

Output Variables
Predicted crime trends and patterns
Crime hotspots and high-risk areas
Insights for law enforcement agencies to inform decision-making and resource allocation

This conceptual framework illustrates the relationships between the input variables, machine
learning model, and output variables, and provides a structure for understanding the crime
prediction system.
 Methodology
This chapter outlines the research methodology used to develop a crime prediction system using
machine learning algorithms. The methodology includes the research design, data collection, data
preprocessing, feature selection, model development, and evaluation.

Research Design
The research design is based on the Knowledge Discovery in Databases (KDD) process, which
involves selecting, preprocessing, transforming, and mining data to discover patterns and
relationships [18]. The KDD process is suitable for this study as it provides a structured approach
to data analysis and knowledge discovery. Research Design
The research design for this study is based on the Knowledge Discovery in Databases (KDD)
process, which is a systematic approach to discovering patterns and relationships in large datasets
[18]. The KDD process consists of the following stages:
1. Selection
This stage involves selecting the relevant data sources and datasets for the study. In this case,
historical crime data from Ndola, Zambia, will be collected and used for analysis.
2. Preprocessing
This stage involves cleaning, transforming, and preparing the data for analysis. Techniques such
as data normalization, feature scaling, and handling missing values will be used to preprocess the
data.
3. Transformation
This stage involves transforming the preprocessed data into a suitable format for analysis.
Techniques such as feature extraction and dimensionality reduction may be used to transform the
data.
4. Mining
This stage involves applying machine learning algorithms to the transformed data to discover
patterns and relationships. Techniques such as classification, clustering, and regression will be
used to analyze the data.
5. Interpretation/Evaluation
This stage involves interpreting and evaluating the results of the analysis. The performance of the
machine learning models will be evaluated using metrics such as accuracy, precision, and recall.
The KDD process is suitable for this study as it provides a structured approach to data analysis and
knowledge discovery. By following the KDD process, this study aims to develop a robust and
reliable crime prediction system that can provide valuable insights for law enforcement agencies.
Data Collection
The dataset used in this study consists of historical crime data collected from various sources,
including:
1. Law Enforcement Agencies
Crime data will be collected from law enforcement agencies in Ndola, Zambia, including crime
reports, incident records, and other relevant documents.
2. Government Websites
Government websites, such as the Zambia Police Service website, will be used to collect crime
data and statistics.
3. Other Sources
Other sources, such as research studies, academic papers, and online databases, may also be used
to collect relevant data.
Dataset Features
The dataset includes the following features:
1. Crime Type: Type of crime committed (e.g., theft, assault, robbery, etc.)
2. Location: Location where the crime was committed (e.g., street address, neighborhood, etc.)
3. Time: Date and time when the crime was committed
4. Socio-Economic Factors: Relevant socio-economic factors, such as poverty rate, unemployment
rate, education level, etc.
Data Characteristics
The dataset is expected to have the following characteristics:
1. Large dataset: The dataset is expected to be large, covering several years of crime data.
2. Multivariate: The dataset includes multiple features, such as crime type, location, time, and
socio-economic factors.
3. Spatial and temporal: The dataset includes spatial and temporal information, such as location
and time of crime.
The collected data will be used to develop and train machine learning models to predict crime
trends and patterns in Ndola, Zambia.

Data Preprocessing
Data preprocessing is a crucial step in the KDD process. The dataset is preprocessed to handle
missing values, outliers, and inconsistencies. Techniques such as data normalization and feature
scaling are used to transform the data into a suitable format for analysis [20].
Feature Selection
Feature selection is the process of selecting the most relevant features that contribute to the
prediction model. Techniques such as correlation analysis and recursive feature elimination are
used to select the most important features [21].
Model Development
The crime prediction model is developed using machine learning algorithms, including Random
Forest and Support Vector Machines. These algorithms are selected for their ability to handle
complex datasets and identify patterns in crime data. The model development process involves the
following steps:
Data Preprocessing
The dataset is preprocessed to handle missing values, outliers, and inconsistencies. Techniques
such as data normalization, feature scaling, and handling missing values are used to preprocess the
data.
Feature Selection
The most relevant features are selected to improve model performance. Techniques such as
correlation analysis and recursive feature elimination are used to select the most important
features.
Model Training
The machine learning models are trained on the preprocessed dataset. The training process
involves optimizing the model parameters to achieve the best performance.
Model Evaluation
The performance of the models is evaluated using metrics such as accuracy, precision, and recall.
The evaluation process involves testing the models on a separate test dataset to assess their
performance.
Model Evaluation Metrics
The following metrics are used to evaluate the performance of the crime prediction model:
Accuracy
The proportion of correctly predicted crime instances. Accuracy is calculated as the ratio of true
positives and true negatives to the total number of instances.
Precision
The proportion of true positives among all predicted positive instances. Precision is calculated as
the ratio of true positives to the sum of true positives and false positives.
Recall
The proportion of true positives among all actual positive instances. Recall is calculated as the
ratio of true positives to the sum of true positives and false negatives.

SDLC Model
The Agile development methodology is used in this study, which involves iterative and
incremental development of the crime prediction system. The Agile methodology allows for
flexibility and adaptability in the development process, enabling the incorporation of changing
requirements and stakeholder feedback. The Agile development process involves the following
stages:
Requirements Gathering
Stakeholder requirements are gathered and documented. The requirements are prioritized and
refined throughout the development process.
Design
The system design is developed and refined. The design stage involves creating a detailed design
document that outlines the system architecture and components.
Implementation
The system is implemented in iterations, with each iteration building on the previous one. The
implementation stage involves coding, testing, and integrating the system components.
Testing
The system is tested and evaluated after each iteration. The testing stage involves verifying that
the system meets the requirements and works as expected.
Deployment
The system is deployed and maintained. The deployment stage involves installing the system in
the production environment and ensuring that it is properly configured and maintained.
The Agile methodology is suitable for this study as it allows for flexibility and adaptability in the
development process, enabling the incorporation of changing requirements and stakeholder
feedback. The iterative and incremental development approach enables the development team to
respond quickly to changing requirements and deliver a high-quality system that meets the
stakeholder needs.
Evaluation
The performance of the crime prediction model is evaluated using metrics such as accuracy,
precision, and recall. The model is also evaluated using techniques such as cross-validation to
ensure its robustness and reliability.

Budget
1. Stationery (pens, notebooks, etc.): K200
2. Transport (fuel, public transport): K800
3. Bundles (data, internet): K500
4. Printing costs (reports, documents): K2000
5. Software/tools and miscellaneous: K1500
Total Estimated Cost: K5000
The total budget of K5000 will be contributed equally by 4 members, with each member
contributing K1250 (K5000 / 4).
Gantt Chart
 References
[1] Y. Wang et al., "Crime Prediction Using Machine Learning: A Survey," IEEE Transactions on
Knowledge and Data Engineering, vol. 31, no. 10, pp. 2439-2454, Oct. 2019.
[2] M. A. Tayebi et al., "Crime Prediction Using Spatial-Temporal Analysis and Machine
Learning," IEEE Transactions on Intelligent Transportation Systems, vol. 20, no. 4, pp. 931-942,
Apr. 2019.
[3] J. Wang et al., "Crime Rate Prediction Using Regression Analysis," IEEE Transactions on
Systems, Man, and Cybernetics, Part C (Applications and Reviews), vol. 42, no. 6, pp. 1456-1465,
Nov. 2012.
[4] Y. Chen et al., "Crime Pattern Analysis Using Clustering Algorithm," IEEE Transactions on
Knowledge and Data Engineering, vol. 25, no. 4, pp. 912-923, Apr. 2013.
[5] K. K. Y. Kuan et al., "Crime Prediction Using Deep Learning: A Review," IEEE Access, vol.
8, pp. 142043-142057, 2020.

[6] M. A. Tayebi et al., "Crime Prediction Using Spatial-Temporal Analysis and Machine
Learning," IEEE Transactions on Intelligent Transportation Systems, vol. 20, no. 4, pp. 931-942,
Apr. 2019.
[7] Y. Zhang et al., "Interpretable Machine Learning for Crime Prediction," IEEE Transactions on
Knowledge and Data Engineering, vol. 32, no. 5, pp. 1234-1245, May 2020.
[8] L. Ngoge, K. Ogada, and D. Kaburu, "Crime Prediction and Mapping Using Machine Learning
Algorithms," Journal of Computing and Information Technology, 2024.
[9] I. H. Sarker, "Machine Learning: Algorithms, Real-World Applications and Research
Directions," SN Computer Science, vol. 2, no. 3, pp. 1-13, 2021.
[10] G. Kanimozhi et al., "Crime Prediction Using Machine Learning," Journal of Intelligent
Information Systems, vol. 58, no. 2, pp. 243-257, 2021.
[11] Pratibha et al., "Crime Prediction Using Machine Learning," International Journal of
Engineering Research & Technology, vol. 9, no. 3, pp. 1-6, 2020.
[12] Llaha, "Crime Prediction and Prevention Using Machine Learning," Journal of Machine
Learning Research, vol. 21, pp. 1-15, 2020.
[13] S. Venkatesh et al., "Crime Prediction Using Machine Learning and Deep Learning," Journal
of Intelligent Information Systems, 2023.
[14] S. Sridharan et al., "Crime Prediction using Machine Learning," EAI Endorsed Transactions
on Internet of Things, vol. 10, no. 39, pp. 1-10, Feb. 2024, doi: 10.4108/eetiot.5123.
[15] U. Fayyad, G. Piatetsky-Shapiro, and P. Smyth, "From data mining to knowledge discovery
in databases," AI Magazine, vol. 17, no. 3, pp. 37-54, 1996.
[16] J. Han, J. Pei, and M. Kamber, Data mining: concepts and techniques. Elsevier, 2012.
[17] I. Guyon and A. Elisseeff, "An introduction to variable and feature selection," Journal of
Machine Learning Research, vol. 3, pp. 1157-1182, 2003.
[18] K. Beck et al., "Manifesto for agile software development," 2001. [Online].
[19] P. J. Brantingham and P. L. Brantingham, "Environmental Criminology," Sage Publications,
1981.
[20] L. E. Cohen and M. Felson, "Social Change and Crime Rate Trends: A Routine Activity
Approach," American Sociological Review, vol. 44, no. 4, pp. 588-608, 1979.
[21] T. M. Mitchell, Machine Learning. McGraw-Hill, 1997.
[22] Parker, Wendy. (2024). Model evaluation. 10.4324/9781003205647-19.

23. Ramkumar PN, Pang M, Polisetty T, Helm JM, Karnuta JM (2022) Meaningless applications
and misguided methodologies in artificial intelligence–related orthopaedic research propagates
hype over hope. Arthroscopy. https:// doi. org/ 10. 1016/j. arthro.
2022. 04. 014
24. Beam AL, Kohane IS (2018) Big data and machine learning in health care. JAMA 319:1317–
1318