Amity Journal of Computational Sciences Volume 7, Issue 1, 2023

ISSN 2456-6616 (Online)

Crime Rate Prediction Using Machine

Abstract: The criminal cases are increasing rapidly in our country policies to reduce crime rates. In this project, we will employ
due to which number of pending cases are also piling up. The a range of ML techniques, including decision trees, random
continuous increase in the criminal cases is proving that it would forests, and support vector machines, to create and evaluate
be very difficult to be classified and to be solved. Recognizing the our predictive model. Data preprocessing methods will be
criminal activity patterns of a place is very important in order to applied to ensure data quality, handle missing values, and
prevent it from happening any other case. The crime solving feature engineering to extract relevant information. In
agencies are doing a better work and can do a better work if they
have a good knowledge of the pattern of criminal activities that are
conclusion, the crime rate prediction system using ML holds
happening in a particular area or region. The most difficult task great promise as a valuable tool in the field of criminology
will most likely be “proving” to legislators that it works. It’s tough and law enforcement. This project aims to contribute to the
to establish the negative when a system is meant to prevent ongoing efforts to improve public safety and reduce crime
something from happening. A positive feedback loop would rates by harnessing the capabilities of machine learning to
certainly benefit companies who are directly involved in providing make accurate predictions and inform evidence-based
governments with ML capabilities to monitor areas or predict decision-making in this critical domain.
crime. Improvements in crime prevention technology will almost
certainly lead to an increase in overall spending on this 1.1 PROBLEM STATEMENT
technology. The proposed system leverages historical crime data,
demographic information, socioeconomic factors, and other The models for the crime prediction methods will be used to
relevant features to train and fine-tune machine learning search through data found in the police archives for data on
algorithms. Various algorithms such as decision trees, random
specific types of crimes as well as the records on different
forests, and support vector machines are employed to analyze and
predict crime rates for specific geographical areas. The project details of their occurrences, with variables that influence the
aims to provide law enforcement agencies with a valuable tool for probability of occurrence studied to get better.
proactive crime prevention and resource allocation, allowing them
to allocate personnel and resources more efficiently in areas with II. PROPOSED ALGORITHM
a higher predicted crime rate. Additionally, the insights gained
from this project can aid policymakers in making informed
decisions to address the underlying socioeconomic factors (a). K-Nearest Neighbours
contributing to crime in different regions.
The K-Nearest neighbours (KNN) algorithm can be utilised
for both regression predictive and classification problems.
I. INTRODUCTION This algorithm suits all considerable parameters. It is usually
utilised for its lower calculation time and ease of
Crime is a persistent and complex societal issue that affects interpretation. There is no need to make additional
communities worldwide, posing significant challenges for assumptions. It works easily on multi-class problems. The
law enforcement agencies and policymakers. Accurate KNN algorithm predicts that alike things subsist near each
prediction of crime rates is crucial for effective resource
other. We can also say, alike things are nearby. For distance
allocation and proactive crime prevention. In recent years, the
calculation, it uses Euclidean, Manhattan, and Minkowski
integration of machine learning (ML) techniques into the
field of criminology has shown promise in enhancing our distance functions. The process is to run the KNN algorithm
ability to forecast crime rates, providing valuable insights for several times with diverse K-value sand select the value of K
law enforcement agencies and policymakers. The that decreases the error number. Using a cross-validation
significance of this project lies in its potential to revolutionize method and by measuring accuracy or validation error, we get
the way we approach crime prevention and public safety. By the optimal value of K.
accurately predicting crime rates at different spatial and
temporal scales, law enforcement agencies can make (b) Decision tree algorithm
informed decisions about resource allocation, personnel
A decision tree is a tree structure where an internal node
deployment, and targeted intervention strategies. Moreover,
depicts a property, the branch refers to a decision rule, and
policymakers can utilize the insights derived from this
predictive model to address underlying socio-economic each leaf node refers to the outcome. The topmost node learns
factors contributing to crime and implement evidence-based to partition depending on the attribute value. It partitions the
tree in a recursive manner called recursive partitioning. This

www.amity.edu/ajcs
 Amity Journal of Computational Sciences Volume 7, Issue 1, 2023
ISSN 2456-6616 (Online)
algorithm aids in decision making. The decision tree concentrate on legitimate crime patterns of crime and
algorithm works in the following way: generate defence strategies, designates areas vulnerable to
criminal action, and then broadcasts security warnings. Our
a) Select the best feature utilising the Gini index or simulation with a large dataset shows that CRIEC can be
cross-entropy to divide the records. much more effective than other models in terms of predicting
b) Makes that feature decision node and splits the set crime. Mary Shermila A, “Crime Data Analysis and
of data into smaller subsets. Prediction of Perpetrator Identity using Machine Learning
Approach [5], The data was/is presented as Prevention is one
of the most prominent and important tasks we have in the
realm of civilization. In addition to being a means of
III. LITERATURE SURVEY identifying and researching the usual patterns and
A literature survey is used to give a succinct but thorough developments of violence, it is also a systemic, scientific
approach. the aim of this model is to make systems more
review of the studies and research articles that have been done
effective at detecting and apprehending criminals This
on a particular topic. The data was/is presented as Crime statistical model can be employed at the crime scenes to
research is a tool used to define criminal activities and study discover crime dynamics and to forecast the description of
them. If the research conducted so far can be seen to be more the criminal most probable perpetrator to be present based on
specifically useful, it is mostly because it indicates which inferences drawn from the site. This is a long, involved
criminal types are useful in controlling crime, then, mostly process of both physically expanding and philosophically
they would it be places where violent crimes are reduced. It advancing. Predictions on who is likely to commit crimes and
is an excellent method for measuring the crime rate because how serious such crimes are the phase includes determining
of each region can be broken down by procedure and the data the number of open offences, which gauges the importance of
is collected for any of each process to be examined. Through different variables, such as the year, month, the weapons
the rapid increase in information technology, crime analysts used, and the social class or demographics of the perpetrators.
will be able to continue to enhance the investigations and help The prediction process is able to deduce how old, whether the
suspect is male, female, and/how many years they've known
them interpret the evidence. on the sample clustering and
the victim There are several theories from the investigation
preprocessing to get unstructured evidence, and then look for
based on the information gathered on this area. The method
crimes inside it. Thus, persons formerly investigated and then uses multilinomial regression, k-neigh regression, and neural
arrested or identified as having committed the same criminal networks for classifiers like Multilearate Regression,
behaviour may often be looked at at for patterns such as kNeighbors, and KNeighbors for static entity definition. The
criminal history, or incident reports, rather than only offences machine learning algorithm was developed and thoroughly
themselves. This is simply intended to direct law enforcement tested using the San Francisco homicide dataset (1981-2014)
resources to where crimes can occur, without attention to and then deployed using Python.
identifying who is responsible. Bayesian classifiers were
used as the current scheme was in use in place. In the current
methodology, the fuzzy C-Means algorithm will be used to IV. IMPLEMENTATION
group the crime data for all items that are apprehensible,
apprehension of, physical assault, larceny-theft, and crime of 4.1. Data Collection
The Crime data set used in this project is in the CSV(Comma
women, as well as all criminal offences such as kidnapping,
separated values) format.
in the dataset.

Nafiz Mahmud, Khalid Ibn Zinna, “CRIMECAST: A Crime 4.2. Data Preprocessing
Prediction and Strategy Direction Service [4], The data was/is The dataset has 10000 entries. df=dropna( ),where df is the
presented as the wide variety of studies on criminology is data frame,is used to remove null values. Label Encoder is
valuable in providing us with a new information on criminal used to translate category attributes(Location,Block,crime
psychology. Criminals don't live in uncertain territory; they types,Community Area)into numeric values. The date
wait before they have an easy target to commit offences, in attribute has been separated into additional characteristics
which case there are clustered areas like hotspots of people such as month and hour, which can be utilized as model
or strangers. It is possible to simulate a crime forecasting features.
model using evidence that can be checked in the fact of past 4.3 Feature Selection
crimes, as long as it has been publicly available, there is The features that will be utilized to build the model are
enough time enough to verify. This paper intends to show chosen. Block, Location, District, Community area, X
how the CCRIMBA's artificial Neural Network has been coordinate, Y coordinate, Latitude, Longitude, Hour and
broadened to include the CRIMAST, a crime prediction and month are the attributes used to pick features.
threat management service which assists law enforcement in
training and testing criminologists to work with Neural 4.4 Building and Training Model
models. the CriMA employs spatial techniques, which The location and month attributes are used for training after
feature selection. The dataset is split into xtrain ,ytrain and x

www.amity.edu/ajcs
 Amity Journal of Computational Sciences Volume 7, Issue 1, 2023
ISSN 2456-6616 (Online)
test and y test pairs. Sklearn is used to import the algorithm not arrest. We used the Chicago crime dataset from 2001 to
model. Models are used to create models a good the present for this prediction [16]. The input data include
fit(xtrain,ytrain). primary kind, local description, beat, district, domestic, ward,
neighborhood, and FBI code. The 10-fold cross-validation
4.5 Prediction Model technique has been used for analysis, hence removing the
predict is used to make predictions after the model has been possibility of overfitting the data. Accuracy, precision, recall,
developed using the above process(xtest). The accuracy is F1 score, and MCC have all been used to gauge performance.
determined by importing accuracy score from metrics- In each example, the split of the data sets was 25% for testing
metrics. accuracy score (y test, predicted). and 75% for training. Tables 5, 6, and 7 display the precision,
recall, and F1 data.
V. RESULTS AND DISCUSSION We can observe from Tables 5, 6, and 7 that the majority of
Different classification methods such as decision tree, K- the examples have precision values between 0.79 and 0.89,
Nearest Neighbour (KNN), and random forest algorithm have with 1.00 being the best. It holds true for recollection as well;
been utilised to forecast different characteristics of crime the majority of recall values fall between 0.44 and 0.58.
data. In this paper, two types of attributes are used to predict Recall has a best value of 1.00. The F1 score ranges from 0.56
crime. One is crime type and the other is the number of to 0.64, with 1.00 being the ideal value. The score for MCC
arrests. The following section discusses the two types of ranges from 0.55 to 0.61. Most of the time, the precision,
attributes in detail. recall, F1 score, and MCC values in our trial findings are not
even near to the best values.We used K (number of nearest
5.1. Prediction Of Crime Types neighbors) = 7 for the KNN algorithm. The Minkowski
Using the KNN, decision tree, and random forest algorithms, distance computation has been applied here in order to
we have predicted different sorts of crimes. Assault, burglary, measure distances.
theft, robbery, weapons violation, car theft, and public peace Using a confidence factor (CF), the decision tree method has
violation are among the several criminal kinds that exist. 632 been applied. When pruning the decision tree, the confidence
instances and 9 attributes have been used in our study. For factor serves as a threshold for permitted inherent inaccuracy
measuring results, a 10-fold cross-validation method has in the data. We choose attributes based on the Gini index
been used. Examining for both training and testing is a criterion. We used CF = 0.45 for all the tests in this research.
benefit of the cross-validation technique. And for the test set, Batch size and iterations have both been employed with the
each observation was used exactly once. In every instance, random forest technique. The batch size (the number of
25% of the data set was split for testing and 75% for training. samples obtained at a time) and iterations for this experiment
Precision, recall, F1 score, and MCC were measured with the are both fixed at 100. We employed several types of numbers
help of the confusion matrix. Table 2, Table 3, and Table 4 (K, CF, or batch size) in these parameters, just as we did with
show the outcomes for KNN and decision tree, and random crime types. However, we went with the number that gives us
forest respectively. the best outcome.
We can observe from Tables 2, 3, and 4 that the majority of
examples have precision values between 0.93 and 1.00, with VII. CONCLUSION
1.00 being the best. It holds true for recall as well; the In conclusion, this minor project on crime rate prediction
majority of recall values fall between 0.94 and 1.00. Recall using machine learning has provided valuable insights into
has a best value of 1.00. The F1 score ranges from 0.92 to the potential of ML techniques in addressing the complex
1.00, with 1.00 being the ideal number. The range of the MCC issue of crime rates in our society. Through a comprehensive
score is 0.92 to 0.99. We can therefore conclude that our exploration of historical crime data, socio-demographic
experimental findings are extremely close to the best values factors, and geographic features, we have developed and
for precision, recall, F1 score, and MCC. We use K (number evaluated predictive models that offer the promise of
of nearest neighbors) = 7 for the KNN method. Using a enhancing law enforcement strategies and public safety
confidence factor (CF), the decision tree method has been measures. While this minor project has made significant
developed. CF is used for pruning. Larger CF gives more progress in the field of crime rate prediction using ML, it is
specific rules to predict the target class. In this paper, we essential to acknowledge that crime prediction is a
applied CF = 0.45. For the random forest algorithm, batch multifaceted issue. Continued research and collaboration
size and number of iterations have been used. For this paper, with law enforcement agencies and policymakers are
both the batch size and the number of iterations are 100. For necessary to refine and deploy these models effectively in
all the cases, we used different types of numbers in these real-world scenarios. As we move forward, it is clear that the
parameters (K, CF, or batch size). But we chose the number integration of machine learning into crime prevention efforts
with which we get the best result. has the potential to make a meaningful impact on public
safety. By leveraging the power of data-driven insights and
5.2. Prediction of the arrest record predictive modeling, we can work toward reducing crime
With the use of the KNN, Decision Tree, and Random Forest rates, improving resource allocation, and ultimately creating
algorithms, we were able to forecast the arrest attribute, safer communities.
which indicates whether or not criminals will be
apprehended. The two goal classifications here are arrest and

www.amity.edu/ajcs
 Amity Journal of Computational Sciences Volume 7, Issue 1, 2023
ISSN 2456-6616 (Online)
VIII. FUTURE ENHANCEMENT [13] R. Kiani, S. Mahdavi, and A. Keshavarzi, “Analysis and
From the encouraging results, we believe that crime data prediction of crimes by clustering and classification”,
mining has a promising future for adding the effectiveness International Journal of Advanced Rese
and effectiveness of felonious and intelligence analysis. arch in Artificial Intelligence (IJARAI), vol.4, no.8, 2015.
[14] PSNI Statistics Branch, Crime-Bulletin [Online]. Available:
Visual and intuitive felonious and intelligence disquisition https://www.psni.police.uk/globalassets/inside-thepsniour-
ways can be developed for crime pattern. As we've applied statistics/police-recorded-crime-statistics [Accessed Jan.13,
clustering fashion of data mining for crime- analysis we can 2019].
also perform other ways of data booby-trapping similar as [15] Kaggle, Chicago Crime Data analysis [online]. Available :
bracket. Also, we can perform analysis on colorful dataset https://www.kaggle.com/gnem/Chicago-crime-rate-analysis
similar as enterprise check dataset, poverty dataset, aid [Accessed Mar.22,2019].
effectiveness dataset, etc. [16] GitHub, Crimes-in-Chicago/crimes in Chicago 2015-2016
[online]. Available: https://github.com/Drachna/Crimes-
IX. REFERENCES inchicago/blob/master/crimes%20in%20chicago%202015-
2016.ipynb [Accessed Mar.20,2019].
[22] P. K. Kushwaha and M. Kumaresan, "Machine learning
[1]. McClendon, Lawrence, and Natarajan Meghanathan. Using
algorithm in healthcare system: A Review," 2021 International
machine knowledge algorithms to anatomize crime data.
Conference on Technological Advancements and Innovations
Machine knowledge and operations An International Journal
(ICTAI), Tashkent, Uzbekistan, 2021, pp. 478-481, doi:
(MLAIJ)2.1 (2015), 1- 12.
10.1109/ICTAI53825.2021.9673220.
[2]. McClendon,L., & Meghanathan,N. (2015). Machine knowledge
[23] P. K. Kushwaha, B. P. Lohani and D. Singh, "Review on
and operations An International Journal (MLAIJ), 2 (1), 1- 12.
information security, laws and ethical issues with online
[3]. Sathyadevan, Shiju. Crime analysis and prophecy using data
financial system," 2016 International Conference on Innovation
mining. 2014 First International Conference on Networks &
and Challenges in Cyber Security (ICICCS-INBUSH), Greater
Soft Computing (ICNSC2014). IEEE, 2014.
Noida, India, 2016, pp. 49-53, doi:
[4]. Sathyadevan, S (2014, August). In 2014 First International
10.1109/ICICCS.2016.7542350.
Conference on Networks & Soft Computing (ICNSC2014) pp.
[24]. G. Gulati, B. P. Lohani and P. K. Kushwaha, “A Novel
406- 412). IEEE.
Application Of IoT In Empowering Women Safety Using GPS
[5]. Kansara, Chirag. ”Crime mitigation at Twitter using Big Data
Tracking Module,” 2020 Research, Innovation, Knowledge
analytics and risk modelling. ”Recent Advances and
Management and Technology Application for Business
Innovations in Engineering(ICRAIE), 2016 International
Sustainability (INBUSH), Greater Noida, India, 2020, pp.
Conference on. IEEE, 2016.
131-137, doi: 10.1109/INBUSH46973.2020.9392193.
[6]. Kim, Suhong, Param Joshi, Parminder Singh Kalsi, and Pooya
[25]. D. Pareta, I. N. Verma, B. P. Lohani, P. K. Kushwaha and V.
Taheri. ”Crime Analysis through Machine Learning.” In 2018
Bibhu, "IoT Enabled Smart and Efficient Musical Water
IEEE 9th Annual Information Technology, Electronics and
Fountain," 2022 2nd International Conference on Innovative
Mobile Communication Conference (IEMCON),pp. 415-420.
Practices in Technology and Management (ICIPTM), Gautam
IEEE, 2018.
Buddha Nagar, India, 2022, pp. 369-373, doi:
[7]. Shah, Riya Rahul. ”Crime Prediction Using Machine
10.1109/ICIPTM54933.2022.9754129.
Learning.” (2003).
[26]. B. P. Lohani, M. Trivedi, R. J. Singh, V. Bibhu, S. Ranjan and P.
[8] N. A. Rahman and W. A. Khader, “KNN classifier and Naive
K. Kushwaha, "Machine Learning Based Model for Prediction
Bayes Classifier for crime prediction in SAN FRANCISCO
of Loan Approval," 2022 3rd International Conference on
context,” International Journal of Database Management
Intelligent Engineering and Management (ICIEM), London,
System(IJDMS), vol.9, no.4, August 2017,
United Kingdom, 2022, pp. 465-470, doi:
doi:10.5121/ijdms.2017.9401.
10.1109/ICIEM54221.2022.9853160.
[9] A. Gupta, A. Mohammad, A. Syed, and M. N. Halgamuge,
[27]. A. Kumar, B. P. Lohani and P. K. Kushwaha, "Robust Secured
“Classification Algorithms using Data Mining: Crime and
Framework for Online Business Transactions over Public
Accident in Denver City the USA”, International Journal of
Network," 2021 2nd International Conference on Intelligent
Advanced Computer Science and Applications
Engineering and Management (ICIEM), London, United
(IJACSA),vol.7, no.7, 2016,
Kingdom, 2021, pp. 555-560, doi:
doi:10.14569/IJACSA.2016.070753.
10.1109/ICIEM51511.2021.9445380.
[10] A. Awal, J.Rabbi, and I.Rana, “Using Data Mining
[28]. P. K. Kushwaha and B. P. Lohani, "A review of security of the
Techniqueto Analyze Crime of Bangladesh”, International
cloud computing over business with implementation," 2016
Journal of Computer Science and Network(IJCSN), vol.6, no.4,
International Conference on Innovation and Challenges in Cyber
ISSN:22175420August 2017.
Security (ICICCS-INBUSH), Greater Noida, India, 2016, pp.
[11] T. Almanie, R. Mirza,and E. Lor, “Crime Prediction based on
192-198, doi: 10.1109/ICICCS.2016.7542342.
crime types and using spatial and temporal criminal hotspots”,
[29]. M. Chandra, P. K. Kushwaha and S. Saxena, "Modified Fractal
International Journal of Data mining & Knowledge
Carpets," 2011 International Conference on Computational
Management Process (IJDKP), vol.5, no.4, July
Intelligence and Communication Networks, Gwalior, India,
2015,doi:10.5121/ijdkp.2015.5401.
2011, pp. 537-540, doi: 10.1109/CICN.2011.115.
[12] P. Yerpude and V. Gudur, “Predictive modelling of crime
[30].P. K. Kushwaha, R. Kohli and D. Singh, "Secret key
dataset using Data mining”, International Journal of Data
watermarking in WAV audio file in perceptual domain," 2015
mining & Knowledge Management Process (IJDKP), vol.7,
International Conference on Futuristic Trends on Computational
no.4, July 2017, doi:10.5121/ijdkp.2017.7404.
Analysis and Knowledge Management (ABLAZE), Greater
Noida, India, 2015, pp. 629-634, doi:
10.1109/ABLAZE.2015.7154940.

www.amity.edu/ajcs
 Amity Journal of Computational Sciences Volume 7, Issue 1, 2023
ISSN 2456-6616 (Online)
[31]. Ranjan, Ankur A. et al. “An Approach for Netflix [44] R. Sharma, M. Mogha, S. Tanwar and A. Rana, "Emerging Part
Recommendation System using Singular Value of Industry 4.0: The Digital and Physical Technology," 2020 9th
Decomposition.” Journal of Computer and Mathematical International Conference System Modeling and Advancement in
Sciences (2019) Research Trends (SMART), Moradabad, India, 2020, pp. 149-
[32]. Makkar, Bhavya et al. “Map Reduce concept-based Sentiment 154, doi: 10.1109/SMART50582.2020.9337064.
Analysis Approach.” International Journal of Computer [45] Dubey, Sanjay Kumar, and Ajay Rana. "Assessment of usability
Sciences and Engineering (2019) metrics for object-oriented software system." ACM SIGSOFT
[33]. Bhatia, Ayush & Bibhu, Vimal & Lohani, Bhanu & Kushwaha, Software Engineering Notes 35.6 (2010): 1-4.
Pradeep. (2020). An Application Framework for Quantum [46] Singh, Archana, Jyoti Agarwal, and Ajay Rana. "Performance
Computing using Artificial intelligence Techniques. 264-269. Measure of Similis and FPGrowth Algo rithm." International
10.1109/INBUSH46973.2020.9392164. Journal of Computer Applications 62.6 (2013): 25-31
[34]. A. Kumar, B. P. Lohani and P. K. Kushwaha, "Black Hole Attack [47] Tyagi, Neha, Ajay Rana, and Vineet Kansal. "Load distribution
in Mobile Ad Hoc Network and its Avoidance," 2021 challenges with virtual computing." Intelligent Computing in
International Conference on Innovative Practices in Technology Engineering: Select Proceedings of RICE 2019. Springer
and Management (ICIPTM), Noida, India, 2021, pp. 103-107, Singapore, 2020.
doi: 10.1109/ICIPTM52218.2021.9388366. [48] Singh, Jaya, and Ajay Rana. "Exploring the big data spectrum."
[35]. Srivastav, A.V., Lohani, B.P., Kushwaha, P.K., Tyagi, S. (2021). International Journal of Emerging Technology and Advanced
Dual-Layer Security and Access System to Prevent the Spread of Engineering 73 (2013).
COVID-19. In: Prateek, M., Singh, T.P., Choudhury, T., Pandey, [49] N. M., P. Chawla and A. Rana, "A Practitioner’s Approach to
H.M., Gia Nhu, N. (eds) Proceedings of International Conference Assess the WCAG 2.0 Website Accessibility Challenges," 2019
on Machine Intelligence and Data Science Applications. Amity International Conference on Artificial Intelligence
Algorithms for Intelligent Systems. Springer, Singapore. (AICAI), Dubai, United Arab Emirates, 2019, pp. 958-966, doi:
https://doi.org/10.1007/978-981-33-4087-9_28 10.1109/AICAI.2019.8701320.
[36]. A. Khuran, B. P. Lohani, V. Bibhu and P. K. Kushwaha, "An AI [50] Tyagi, N., Rana, A., Awasthi, S., & Tyagi, L. K. (2022). Data
Integrated Face Detection System for Biometric Attendance Science: Concern for Credit Card Scam with Artificial
Management," 2021 2nd International Conference on Intelligent Intelligence. In Cyber Security in Intelligent Computing and
Engineering and Management (ICIEM), London, United Communications (pp. 115-128). Singapore: Springer Singapore.
Kingdom, 2021, pp. 29-33, doi: [51] Jain, Piyush, Sanjay Kumar Dubey, and Ajay Rana. "Software
10.1109/ICIEM51511.2021.9445295. usability evaluation method." International Journal of Advanced
[37]. S. Salagrama, B. P. Lohani and P. K. Kushwaha, "An Analytical Research in Computer Engineering & Technology 1.2 (2012): 28-
Survey of User Privacy on Social Media Platform," 2021 33.
International Conference on Technological Advancements and [52] Pal, S. K., et al. "Hanging suicides in himachal pradesh: an
Innovations (ICTAI), Tashkent, Uzbekistan, 2021, pp. 173-176, analysis of forensic cases." Int J Forensic Sci Pathol 4.11 (2016):
doi: 10.1109/ICTAI53825.2021.9673402. 297-304.
[38]. S. Singh, D. Chaudhary, A. D. Gupta, B. Prakash Lohani, P. K. [53] Rana, A., and S. Manhas. "Significance of diatoms in diagnosis
Kushwaha and V. Bibhu, "Artificial Intelligence, Cognitive of drowning deaths: a review." Journal of Forensic & Genetic
Robotics and Nature of Consciousness," 2022 3rd International Sciences 5 (2018): 77-81.
Conference on Intelligent Engineering and Management [54] Bansal, Sangeeta, and Dr Ajay Rana. "Transitioning from
(ICIEM), London, United Kingdom, 2022, pp. 447-454, doi: relational databases to big data." International Journal of
10.1109/ICIEM54221.2022.9853081. Advanced Research in Computer Science and Software
[39]. S. Suman, P. Kaushik, S. S. N. Challapalli, B. P. Lohani, P. Engineering 4.1 (2014): 394-400.
Kushwaha and A. D. Gupta, "Commodity Price Prediction for
making informed Decisions while trading using Long Short-Term
Memory (LSTM) Algorithm," 2022 5th International Conference
on Contemporary Computing and Informatics (IC3I), Uttar
Pradesh, India, 2022, pp. 406-411, doi:
10.1109/IC3I56241.2022.10072626.
[40] P. William, Y. V. U. Kiran, A. Rana, D. Gangodkar, I. Khan and
K. Ashutosh, "Design and Implementation of IoT based
Framework for Air Quality Sensing and Monitoring," 2022 2nd
International Conference on Technological Advancements in
Computational Sciences (ICTACS), Tashkent, Uzbekistan, 2022,
pp. 197-200, doi: 10.1109/ICTACS56270.2022.9988646.
[41] Mridul Bhardwaj and Ajay Rana. 2015. Impact of Size and
Productivity on Testing and Rework Efforts for Web-based
Development Projects. SIGSOFT Softw. Eng. Notes 40, 2 (March
2015), 1–4. https://doi.org/10.1145/2735399.2735404
[42] Bhardwaj, Mridul, and Rana Ajay. "Estimation of testing and
rework efforts for software development projects." Asian Journal
of Computer Science and Information Technology 5.5 (2015): 33-
37.
[43] Dubey, Gaurav, Ajay Rana, and Jayanthi Ranjan. "A research
study of sentiment analysis and various techniques of sentiment
classification." International Journal of Data Analysis Techniques
and Strategies 8.2 (2016): 122-142.

www.amity.edu/ajcs