ROLE OF CRIME PREDICTIVE TECHNIQUES IN THE INVESTIGATION OF CYBERCRIMES IN INDIA
Cybercrimes threats are growing, and with this, the need to use distinct investigative techniques for its investigation. However, cybercrime investigating agencies, especially in India, use traditional investigative techniques similar to the investigating techniques in a traditional crime. With the advent of technological augment, there is a necessity of new techniques for investigation possibly to predict and prevent cybercrimes, for example, through the use of crime predictive techniques. There are various models where researchers have classified cybercrimes data using crime predictive techniques and have been able to achieve an accuracy rate of 99%. The enforcement agencies in Maharashtra are one of the first to adapt to the ways of predictive policing by inculcating these models in the process of investigation. But there are researchers and advocates of human rights who oppose the use of such techniques because they believe that these techniques institutionalize discrimination and are violative of certain human rights. Further, there are research studies and examples which suggest that predictive techniques used for policing are ineffective. The present research paper, in contrast to this, has tried to show that crime predictive techniques are still in the stage of development. With new methods of commission of cybercrimes being invented almost every day, predictive techniques need to be as fast-paced. The paper recommends that the enforcement agencies with collaborations with research institutions in this field should come up with solutions that are neither violative of human rights nor so inaccurate that it is rendered ineffective to combat cybercrimes effectively.
Investigation of cybercrimes in India
Investigation can be termed as the dawn of a criminal trial. It usually commences with the registration of the First Information Report (“FIR”), wherein the investigating agencies investigate the veracity of the complaint or information received. The procedure of investigation often changes with the nature of the offence committed and hence, for offences like cybercrimes, criminalized and penalized under the Indian penal Code, 1860 (“IPC”) and the Information Technology Act, 2000 (“IT Act”), technological know-how becomes an important investigative tool. The specific techniques used in the investigation of cybercrimes are complex enough that many enforcement agencies, including the Central Bureau of Investigation (“CBI”), have created distinct cells for the investigation and handling of cyber space offences. Bangalore, being the Information Technology or IT hub of India, was the first city to establish a Cyber Crime Police Station.
To further ease the process of investigation in cyber-crimes, NASSCOM and DSCI, representing the Indian IT industry, under an exclusive partnership, came up with a Cyber Crimes Investigation Manual in 2011. A need for such collaboration and creation of the manual was felt because of the rising menace of cybercrimes. As of 2021, more than 3.17 lakh cybercrimes have been registered online since the inception of the IT Act. With the growing threat of cyber space offences the need for distinct investigative techniques is becoming acute to prevent as well as to detect these crimes.
However, before delving into the investigative techniques, one needs to understand the scope of cybercrimes. Cybercrimes are inclusive of all acts or commissions done with criminal intent in cyberspace. Cybercrimes, majorly in India, are reported under the category of computer hacking, forgery, counterfeiting, publication or transmission of pornographic or obscene content, and fraud. The primary offences dealt with within the IT Act are tampering with source code, deleting or altering data with malicious intent, and transmitting or publishing pornographic material. Nevertheless, cybercrimes are not limited to only these categories of offences. Certain other offences added in the IT Act through subsequent amendment are cyber-stalking, worm attack, cybersquatting, identity theft, financial crimes, web jacking, pharming, spamming, skimming, website defacement, email spoofing, cyber terrorism, acts distressing the integrity and sovereignty of India, etc. The IT Amendment Act, along with the amendments to IPC and Indian Evidence Act, 1872 (“Evidence Act”), paved the way for prosecution of offences like phishing and other financial crimes. In spite of this, the trial and prosecution of cybercrimes are not resorted to due to a lack of awareness of the applicability of these general laws to cybercrimes.
But now the question arises, how are cybercrimes committed? And, is there a special procedure for its investigation? Various techniques and technological tools are used in the commission of cybercrimes such as buffer overflow, phishing, malware, salami attack, data diddling, cracking, steganography, trojan, zombie, etc. Investigation of cybercrimes due to the use of such technical tools is primarily done through the gathering of digital evidence and cyber forensic analysis of the digital evidence. Cyber forensics can be further classified into disk forensics, wireless forensics, network forensics, malware forensics, email forensics, database forensics, SatNav forensics, and memory forensics. The cyber forensics process requires identification of digital evidence, preservation of evidence, analysis of evidence, reporting of findings, and adducing of digital evidence before the trial court. Moreover, cyber forensic analysis happens subsequent to the examination of any possible alteration, data corruption or virus introduction, the discovery of files on the subject system, recovery of deleted or hidden files, access to protected or encrypted files, analysis of all relevant data, analysis of the timeline of events occurred, etc.
These techniques are traditional methods of investigation used in cybercrimes. Cybercrime investigating agencies, especially in India, use these traditional investigative techniques, particularly concerning information and evidence gathering. Furthermore, the old method of interviewing suspects, witnesses, and victims is still followed during the investigation of cybercrimes in India. With the advent of technological augment, there is a necessity of innovative techniques for investigation, possibly to prevent even the happening of such crimes. The traditional techniques of detection of cybercrimes, as of now, seem to be incapable of entirely stopping or mitigating these crimes. This is because the target groups or the victims differ for different cyberspace offences and the cyber criminals improvise their methods of attacks with the advancement in technology.
Review of Literature
Before proceeding further, it is imperative to review the following literature on the investigating techniques used in cybercrimes:
In the article Analysis of Cybercrime Investigation Mechanism in India, the authors M. Elavarasi and N. M. Elango state that the process of cybercrime investigation includes questioning information gathering and computer forensics. As per the authors, the techniques of investigation are searching the offender, tracking the IP address, analyzing the webserver, tracking the emails, recovery of deleted data, ethically hacking the system for the purpose of investigation, and uncovering hidden data on the subject system. This article, however, does not probe into the mechanics of cyber forensics. It simply states the steps involved in the process of investigation of cybercrimes without elaborating on the crime detection methods.
In the article Research on Cybercrime and its Policing, published in the American Journal of Computer Science and Engineering Survey, the authors have emphasized the role of investigating agencies in curbing cybercrimes. The authors state that because of lack of resources, under-reporting, and low reporting rate, the chances of prosecution of offenses committed in cyberspace are quite low. The authors identify training of the personnel as a major challenge when it comes to such digital crimes. But the investigating agencies have formed partnerships with the computing divisions of universities for ease of investigation. Moreover, the authors suggest collaboration with private companies as they have more expertise and experience in dealing with such cases. Lastly, the authors recommend the use of Artificial Intelligence for the identification of threats whereby the problem gets detected in the openness in a short span of time.
But what are these techniques used in the detection of threats or cybercrimes? As per the article Comprehensive Review of Cybercrime Detection Techniques, there are various new models of crime detection and prediction techniques. A few of them are as follows: statistical, machine learning, data mining, and other techniques such as biometric, cryptography, or computer visions. The paper does not differentiate between detection and predictive techniques. These two are mostly grouped together.
However, prediction is a different step when it comes to the analysis of cybercrimes data. Ensar Seker states that the present focus should be on prediction systems that can anticipate complex cyber-attacks rather than being on the defence mechanism and trying to mitigate the damages incurred. Similarly, the author of the article Computational System to Classify Cyber Crime Offenses using Machine Learning emphasizes the predictive techniques by stating that while analyzing data in the prediction step, the cybercrime data is used to predict which crimes are occurring in larger number in a particular year at a particular location. Such analysis helps the investigating agencies to reduce the manifestation of future cybercrime incidents.
Furthermore, Akshay Kumar Singh, Neha Prasad, Nohil Narkhede, and Siddharth Mehta, in their article Crime: Classification and Prediction Pattern, have defined a structure for analyzing crime and have also given a discourse on the method of increasing the accuracy of crime prediction for the deterrence of crimes. They have used the method of ‘A Priori Algorithm’ for the identification of drifts and patterns in crime. Their method of prediction divides the crime into classes or categories of crime. Hence, it is made to aid the investigating agencies in the employment of measures for the prevention of crimes. It can also predict the timeline of the crimes if the data collected is complete and comprehensive.
Few other authors, for example, in the article Crime Pattern Analysis, Visualization and Prediction Using Data Mining, have also grouped crime data together according to different types of crimes against women that had taken place specifically in various cities of India. The authors in this article have used various statistical predictive methods to cluster, correlate crimes, and finally make a crime prediction. In the article Survey of Analysis of Crime Detection Techniques Using Data Mining and Machine Learning, the authors mentioned different machine learning and data mining techniques while analyzing their work. They have reviewed various types of crimes, including cybercrimes. They have used various techniques such as hidden Markov model, genetic algorithm, neural network, logistic regression, kernel destiny algorithm, k-means, and random forest.
Taking into consideration the above-reviewed literature, the present research paper aims to understand the crime predictive techniques used in cybercrime investigation. The paper will, in the upcoming sections, study the crime predictive techniques used by different countries in the investigation of cybercrimes and, in light of that, try to understand the methods used in India. Lastly, the paper will try to come up with proposals, if any, for efficient investigation of cybercrimes in India through the use of crime predictive techniques.
What are ‘Crime Predictive’ Techniques?
Crime predictive techniques or also known as predictive policing, are not a new method of investigation. It involves the use of algorithms to analyze huge data sets in order to predict and prevent the commission of future crimes. One of the first countries to adopt such methods of investigation was the United States of America, specifically the Los Angeles Police Department in 2008. They have implemented such policing methods wherein the areas with the likelihood of gun violence are identified, and hot-spots for certain other property-related crimes are intensified when it comes to patrolling.
There is also evidence of the use of such crime predictive techniques in the investigation of cybercrimes. However, before digging into the predictive techniques, it is important to ask the question that are detection and predictive techniques the same? On a survey of existing literature, no such difference was observed. The techniques used for crime detection are used for crime prediction as well. Therefore, some of the latest crime detection or predictive techniques used are as follows:
1. Statistical: Statistical method is used mostly for evaluating and extracting data for the effective detection of cyberattacks. This method is carried out by using the K-means clustering in a certain grouping of crimes. K-means is a method in which the data can belong to more than one group. Statistical method is also used in machine learning techniques sometimes for predicting outcomes. However, due to lack of evidence, it can also happen that the offender’s behaviour is considered to understand whether he has committed some other undetected crime in this method.
2. Machine Learning: It is the method of predicting outputs constructed on input data fed into the machine, i.e., the computer. The data fed into the machine is also called training data. Machine learning process can be either unsupervised or supervised. In the unsupervised method, the machine does not know the right output for each input data. However, in supervised machine learning, the input data and corresponding data output are paired, and hence, the correct data is shown as output. A neural network is also a form of machine learning, and it works in a similar fashion to that of the neurons in the brain. Neural learning can be further sub-divided into deep learning and fuzzy logic neural networks.
3. Data Mining: Data Mining uses the ‘A Priori’ algorithm for the detection/prediction of cybercrimes. Usually, cases are identified from which variables of the cases are determined and extracted. Thereafter, visual representations are made, for example, graphs, bar charts, etc., for easier comprehension of the analysis by the investigating agencies.
4. Geographical System: Crime hot-spots are identified, for example, a shopping complex, etc. Identification of such crime generators is based on the idea that the previous victimization is a good predictor of upcoming risks.
5. Other Techniques: Some of the techniques that are included in this category of crime prediction techniques are biometric, forensic tools, cryptography, and computer vision. There are also other techniques such as detection of deviation, string comparator, and social network analysis.
These are not the only techniques used for crime prediction. There are several algorithms that are used to produce accurate results using other techniques, for example, decision tree, KNN classification, Bayesian method, Logistic Regression, Random Forest, etc. The above-mentioned techniques are the most commonly used techniques across different countries.
Using Crime Predictive Techniques in Investigation of Cybercrimes
The internet has become a potent weapon in this century. With new innovations happening every day, cyber crimes have become a hazard that requires quick solutions. Traditional investigating methods take a longer time to predict plausible offenders, future crime-scene or even to understand the pattern of the cybercrimes, and hence, it is necessary these days to analyze the crime patterns, crime-scene, criminals, etc. using new techniques or methods.
One of the major concerns in the investigation of cybercrimes is that the investigating agencies are not adequately trained to be able to investigate cyberspace cases. The workforce is becoming deficient as technology progresses with time, and therefore, even new resolutions are falling short. But, it is the duty of the enforcement agencies to ensure that cybercrimes are effectively investigated. Researchers, with the help of crime prediction techniques and analysis of data, can help these agencies in fighting cybercrimes.
A few examples of such crime prediction in traditional crimes are risk mapping, criminal detection, etc. On carrying out predictive analysis on murders that occurred in ten years in Brazil, the accuracy rate shown was 97%. Random Forest method was used in order to understand the effect of urban metrics on murders. Random forest is an algorithm that creates a classifier on training data through a combination of outputs to have the best possible predictions on test data. Ignorance and unemployment were identified as important variables in the analysis of murders in Brazil. The study also highlighted the pressing need for predictive analysis of crime data for ease of investigation.
Attempts, through predictive analysis, have also been made to combat cybercrimes; for example, a model was built to predict phishing attacks, and satisfactory accurate results were observed. Also, there are surveys to show that machine learning techniques have been effective in tackling the problem of ‘online sexual grooming.’
One recent study had taken five years of cybercrimes data of Turkey for predictive analysis.  Through the use of machine learning techniques, the mode of attack and offender were predicted. Factors such as age, income, education, marital status, etc., were considered for the analysis. These factors were helpful in envisaging the mode of attack and victims of these cybercrimes. The accuracy rate of predicting the offender in these cybercrimes was found to be around 60%. The researchers in this study tried to link the victim’s education and income level to that of the occurrence rate. They found that the occurrence rate is inversely proportional to the victim’s educational and income level. The objective of this study was to aid the enforcement agencies and to provide a faster mechanism for predicting criminals.
Investigation of Cybercrimes through Crime predictive techniques in India
There have been few studies conducted in India to analyze traditional crimes as well as cybercrimes. In one such study, using KNN (machine learning) method, the location and time of the crimes were used to conduct a predictive examination of crimes in certain parts of India. This study predicted the place and time of the crimes like murder, robbery, accident, gambling, kidnapping, and violence. Prediction of terrorist activities has also been attempted through analysis of geographical and demographic events that happened in the past years. The success rate of prediction in this particular study was impressive.
To understand the use of predictive techniques in India better, the following studies have been relied on:
Machine Learning Analysis to classify Cybercrime Offenses: Different techniques of machine learning were used to analyze data set collected from all over India. The data set contained information about the type of cybercrimes, age of the offenders, victims, access violation, year and place of occurrence, and damage incurred by the victim. For classification of the crimes, the Bayesian method and for clustering K-means were used. This model, as per the researchers, classified data with 99% accuracy. The model saved the amount of time that would have been consumed in manual reporting and classification of these cybercrimes. However, changes will have to be made for the investigating agencies to be able to curb the growth of cybercrimes in a specific area/location.
Cybercrime Detection on social media using Random Forest technique: This study was based on the premise that for ease of cybercrime investigation, data mining techniques can be used. The researchers in this study have collected data from the social media interface, which is then classified using algorithms. The parameters used in the analysis of data were age, location, synonyms, hashtags, gender, and keywords. Random Forest method was found to be most accurate in the parameter-based detection of threats. It showed an accuracy rate of 80%.
When it comes to the implementation of such research or studies, only a few initiatives can be seen. Gradually, the investigating agencies, especially the police, in Maharashtra is adapting to the ways of predictive policing, which happens to be one of the most digitally progressive forms of monitoring by the police of the future. There is evidence where cybercrime experts have admitted to the fact that predictive techniques will be the most effective tools for investigation. Around 47 cybercrime labs have been established in Maharashtra. Also, the Government is trying to establish cybercrime police stations for the facilitation of faster registration of cybercrimes and forensic investigation of such digital crimes.
The Jharkhand Police, in 2013, in partnership with the National Informatics Centre, started a project for the development of a data mining software that will analyze the data to trace the crime trends and patterns. Delhi Police has also, in collaboration with the Indian Space Research Organisation (“ISRO”), developed a predictive tool for policing called ‘Crime Mapping, Analytics and Predictive System’ or CMAPS. The system of CMAPS identifies hotspots of crimes by the combination of calls placed in the Delhi helpline number and satellite imagery provided by ISRO’s that basically performs the function of clustering. After the use of CMAPS, Delhi Police has reduced the time taken to analyze a crime, on an average 15 days, to that of only three minutes. Even though these are examples of traditional crime predictive mechanisms, it showcases that India is trying to keep up with the trends of the advancing technology and hence, would perhaps in the future implement predictive techniques with more rigour for the investigation of cybercrimes.
Are Crime predictive techniques Violative of Fundamental Rights?
Crime predictive techniques, in the above sections, were portrayed with a virtuous stance. However, there are researchers and advocates of human rights who oppose the use of such techniques because they believe that these techniques institutionalize discrimination and are violative of certain fundamental rights.
There is ample evidence to suggest that the police in India are casteist, discriminatory, and communal. Religious minorities and lower castes have been historically the victims of violence in the hands of the system. In the case of Ankush Maruti Shinde v. the State of Maharashtra, the Supreme Court, in review petitions, acquitted these six men who had been on death row for almost sixteen years, and that too in solitary confinement. These men belonged to a marginalized community and were apprehended by the police as offenders even when an eyewitness had identified four other men as the actual offenders. Not only this, but marginalized communities like Dalits, tribals, and Muslims who constitute as per the 2011 census 16.6%, 8.6%, and 14.2% of the population constitute around three-fourth of the undertrial prisoners’ population in the prisons. To be precise, out of the total undertrial prisoners in India, 21.6% are Dalits, 11.8% are tribals, and 19.7% are Muslims. Similarly, in the United States, Blacks only constitute 12% of the population, whereas they account for around 33% of the inmate population.
Prediction of similar discrimination in predictive policing is being made by a few scholars. In fact, scholars in the United States of America have urged other researchers working in the field of predictive policing to stop working on the same as they believe it purports structural or institutionalized racism. Researchers have also tried to demonstrate the same bias being enforced against the marginalized communities in the predictive techniques in India. A particular study shows that CMAPS, initiated by the Delhi police, is discriminatory in nature. Hence, criminal predictive techniques are perceived to be another tool that will institutionalize such discrimination. The way this discrimination is envisaged is that the data fed into the algorithms would be discriminatory in nature, and therefore, the results, in turn, would be biased. In simpler terms, it “isn’t necessarily reflective of who is more likely to commit a crime; rather, it is an indicator of who is more policed.”
On paper the predictive algorithms used for policing seem to be beneficial but in reality, the algorithms target only a certain group or community as the people feeding data sets to the machine are biased. This leads to the production of biased results and consequently, only the people who were always at a disadvantage, to begin with, are apprehended. Moreover, the constitutional validity of these algorithms can be challenged in light of cases such as Lt. Col. Nitisha v. Union of India and Madhu v. Northern Railways, which have held that policies that are neutral on paper but discriminatory in practice can violate the right against discrimination (systemic discrimination).
However, all of these concerns are with regards to traditional crimes’ investigation. There aren’t pieces of evidence of such discrimination when it comes to cybercrimes per se, but these arguments possibly could also be made against the use of predictive techniques in the investigation of cybercrimes. Lack of accuracy, accountability, and privacy concerns are other arguments against the use of predictive techniques in cybercrimes investigation.
The usage of data available online, on social media, or with the government for investigation of cybercrimes raises questions about violation of one’s privacy. Enforcement agencies might argue that since these were voluntarily given, the issue of privacy cannot be raised. But, can it be said that everything that is online can be used to incriminate an individual? Few lawyers view this as a violation of one’s privacy which is neither legally nor morally right. Moreover, Rule 3(5) of the Intermediary Rules 2018, which regulates the transmission of information on the internet, enables the intermediaries to be able to trace the originators of info on the platform. These companies could be pressured to monitor data and hand it over whenever the enforcement agencies or the government desires. After the Puttaswamy judgment, these techniques could be challenged as being violative of Article 21 on the ground of violation of privacy.
Are Crime Predictive Techniques Effective?
There are research studies and examples which suggest that predictive techniques used for policing are ineffective. The Chicago police used predictive algorithms for the identification of around 400 people who were at the highest risk of being victims of murders in the year 2013-14. Later, it was observed that only three of those identified as potential victims were on the list of people who were actually homicide victims. In Florida, the incorporation of these predictive techniques led to an increase in property crimes. Predictive policing systems sold by a company in this industry called PredPol were discontinued in the United States as they were ineffective. Not just United States but also in countries like Berlin, these techniques were, after a trial, declared ineffective. The reason for such failure is when data becomes the sole identifier, and the enforcement agencies fail to understand the cause-and-effect relationship between the data fed and the output obtained. For example, if it is believed that the administration of drugs is a major reason for the occurrence of a lot of violent crimes, then this is just correlation and not causation. The algorithms seem unable to identify this gap in their prediction mechanism.
Also, in the United Kingdom, the National Crime Agency has been identifying ‘at-risk’ children based on their online activity/presence, which then becomes indicative of potential interest in “cybercrime forums or the purchase of cybercrime tools.”  This puts the children or youngsters on the radar of the Crime Agency even before they have engaged in any criminal activity online based on mere web-engine searches, which might seem absurd.
With technology, there will be quite a few instances of failure as they are at the end made by humans, and humans are susceptible to error. But there are researchers and enforcement officers who firmly believe that predictive techniques will result in wondrous control of cybercrimes. The cases studies elaborated in this research paper also provide high rates of accuracy, which can be achieved if the data fed into the machine is unbiased, free from prejudices, and the techniques are updated regularly, keeping in view the current technology used in the commission of cybercrimes.
Technology, when used with malice, will result in unfavorable outputs. But technology, when used without pre-set prejudices and biases and in a system that is against the idea of discrimination, would probably never prey upon the weaker communities. Discrimination ingrained in the system in India cannot be obliterated overnight. It has been an ongoing issue that either has to be sidestepped or checked through various checkpoints. If predictive techniques are stopped from being used in the investigation, in my view, it would not really address the issue of discrimination. The enforcement agencies will continue to target the marginalized communities. However, if the predictive techniques are designed in a manner that addresses the issue of discrimination or misuse in the investigation of not just cybercrimes but also traditional crimes, then these techniques can be put to use.
Crime predictive techniques are still in the stage of development. With new methods of commission of cybercrimes being invented almost every day, predictive techniques would need to be as fast-paced. The enforcement agencies with collaborations with research institutions in this field will have to come up with solutions that are neither invasive of one’s privacy nor so inaccurate that it is rendered ineffective.
[The author is a L.L.M. student at NALSAR University, Hyderabad]
 Niryat Bhawan & Rao Tula Ram Marg, Cyber Crime Investigation Manual (2011).
 Id. at 10.
 Id. at 5.
 3.17 lakh cybercrimes in India in just 18 months, says govt, The Hindu, March 9, 2021, https://www.thehindu.com/sci-tech/technology/317-lakhs-cybercrimes-in-india-in-just-18-months-says-govt/article34027225.ece (last visited Nov 7, 2021).
 Abhijeet Deb, Cyber Crime and Judicial Response in India, 3 Indian J. Law Justice 106–117 (2012).
 Bhawan and Marg, supra note 1 at 9.
 N. S. Nappinai, Cyber Crime Law in India: Has Law Kept Pace with Engineering Trends - An Empirical Study, 5 J. Int. Commer. Law Technol. 22–28 (2010).
 The Information Technology (Amendment) Act, 2008.
 Nappinai, supra note 7 at 22-23.
 Id. at 23.
 Bhawan and Marg, supra note 1 at 12-13.
 Id. at 20.
 Id. at 21.
 Dorothy A Lunn, Computer Forensics - An Overview 8 (2000).
 katharina.kiener-manu, Cybercrime Module 6 Key Issues: Handling of Digital Evidence, //www.unodc.org (last visited Nov 7, 2021).
 Satish Kumar Ganta, CYBER CRIMES IN INDIA – AN INVESTIGATION PERSPECTIVE 5.
 Wadha Abdullah Al-Khater et al., Comprehensive Review of Cybercrime Detection Techniques, 8 IEEE Access 137293–137311 (2020).
 Id. at 137296.
 M. Elavarasi et al., Analysis of Cybercrime Investigation Mechanism in India, 10 Indian J. Sci. Technol. 1–4 (2017).
 Dhaval Chudasama et al., Research on Cybercrime and its Policing, 8 14 (2020).
 Id. at 2.
 Al-Khater et al., supra note 17.
 Id. at 137297.
 Ensar Seker, Attack Pattern Detection and Prediction, Medium (2020), https://towardsdatascience.com/attack-pattern-detection-and-prediction-108fc3d47f03 (last visited Nov 7, 2021).
 Rupa Ch et al., Computational System to Classify Cyber Crime Offenses using Machine Learning, 12 Sustainability 4087 (2020).
 Akshay Ku. Singh et al., Crime: Classification and Pattern Prediction, 3(2) IARJSET 41-43 (2016).
 Tushar Sonawane et al., Crime Pattern Analysis, Visualization And Prediction Using Data Mining, 1 Int. J. Adv. Res. Innov. Ideas Educ. 681–686 (2015).
 S. Prabakaran & Shilpa Mitra, Survey of Analysis of Crime Detection Techniques Using Data Mining and Machine Learning, 1000 J. Phys.: Conf. Ser. 012046 (2018).
 179, Predictive Policing Explained | Brennan Center for Justice (2020), https://www.brennancenter.org/our-work/research-reports/predictive-policing-explained (last visited Nov 7, 2021).
 Al-Khater et al., supra note 17 at 137297.
 Farhat Bibi & Dr Hamid Ghous, Review On Cybercrime Prediction with Data Mining Techniques, 79 21 (2021).
 M Premasundari & C Yamini, A Violent Crime Analysis Using Fuzzy C-Means Clustering Approach, 09 ICTACT J. SOFT Comput. 6 (2019).
 Bibi and Ghous, supra note 32 at 1.
 Vikas Grover, Richard Adderley & Max Bramer, Review of Current Crime Prediction Techniques 233–237 (2007).
 Al-Khater et al., supra note 17 at 137298-299.
 Id. at 137300.
 Id. at 137302-303.
 Id. at 137304.
 Grover, Adderley, and Bramer, supra note 36.
 Al-Khater et al., supra note 17. at 137306.
 Grover, Adderley, and Bramer, supra note 36.
 Al-Khater et al., supra note 17.
 Pranay Jha, Raman Jha & Ashok Sharma, Behavior Analysis and Crime Prediction using Big Data and Machine Learning, 8 8 (2019).
 Chudasama et al., supra note 21 at 2.
 Abdulkadir Bilen & Ahmet Bedri Özer, Cyber-attack method and perpetrator prediction using machine learning algorithms, 7 PeerJ Comput. Sci. e475 (2021).
 Andrew P. Wheeler & Wouter Steenbeek, Mapping the Risk Terrain for Crime Using Machine Learning, 37 J. Quant. Criminol. 445–480 (2021).
 S. T. Bharathi, B. Indrani & M. Amutha Prabakar, A supervised learning approach for criminal identification using similarity measures and K-Medoids clustering, in 2017 International Conference on Intelligent Computing, Instrumentation and Control Technologies (ICICICT) 646–653 (2017).
 Luiz G. A. Alves, Haroldo V. Ribeiro & Francisco A. Rodrigues, Crime prediction through urban metrics and statistical learning, 505 Phys. Stat. Mech. Its Appl. 435–443 (2018).
 Bilen and Özer, supra note 47.
 Rajalakshmi Shenbaga Moorthy & P. Pabitha, Optimal Detection of Phising Attack using SCA based K-NN, 171 Procedia Comput. Sci. 1716–1725 (2020).
 C.H Ngejane et al., Mitigating Online Sexual Grooming Cybercrime on Social Media Using Machine Learning: A Desktop Survey, in 2018 International Conference on Advances in Big Data, Computing and Data Communication Systems (icABCD) 1–6 (2018), https://ieeexplore.ieee.org/document/8465413/ (last visited Nov 8, 2021).
 Bilen and Özer, supra note 47.
 Akash Kumar et al., Crime Prediction Using K-Nearest Neighboring Algorithm, in 2020 International Conference on Emerging Trends in Information Technology and Engineering (ic-ETITE) 1–4 (2020).
 Dinesh Verma et al., Understanding Patterns of Terrorism in India (2007-2017) Using Artificial Intelligence Machine Learning, 15 Int. J. Technol. Knowl. Soc. 23–39 (2019).
 Ch et al., supra note 26.
 Twinkle Arora, Monika Sharma & Sunil Kumar Khatri, Detection of Cyber Crime on Social Media using Random Forest Algorithm, in 2019 2nd International Conference on Power Energy, Environment and Intelligent Control (PEEIC) 47–51 (2019), https://ieeexplore.ieee.org/document/8976474/ (last visited Oct 25, 2021).
 Data mining to drive predictive policing in the future of Maharashtra police, , Hindustan Times (2017), https://www.hindustantimes.com/pune-news/data-mining-to-drive-predictive-policing-in-the-future-of-maharashtra-police/story-GIl9jy7W0AJYBAPzhy9dSO.html (last visited Nov 8, 2021).
 Predictive Policing: What is it, How it works, and its Legal Implications — The Centre for Internet and Society, , https://cis-india.org/internet-governance/blog/predictive-policing-what-is-it-how-it-works-and-it-legal-implications (last visited Nov 9, 2021).
 Gaurav Jain & Raghav Chopra, AI tech is increasingly being used by police worldwide. Here’s why India needs to regulate it, Scroll.in , https://scroll.in/article/989094/ai-tech-is-increasingly-being-used-by-police-worldwide-heres-why-india-needs-to-regulate-it (last visited Nov 8, 2021).
 The Police in India Is Both Casteist and Communal, , The Wire , https://thewire.in/caste/police-casteist-communal (last visited Nov 9, 2021).
 Hidden Apartheid: Caste Discrimination against India’s “Untouchables,” (2007), https://www.hrw.org/report/2007/02/12/hidden-apartheid/caste-discrimination-against-indias-untouchables (last visited Nov 9, 2021).
 Ankush Maruti Shinde v. State of Maharashtra (2019) 15 S.C.C. 470 (India).
 In a minority, but a major presence in our prisons - Times of India, , The Times of India , https://timesofindia.indiatimes.com/india/in-a-minority-but-a-major-presence-in-our-prisons/articleshow/73266299.cms (last visited Nov 9, 2021).
 Courtney Linder, Why Hundreds of Mathematicians Are Boycotting Predictive Policing, Popular Mechanics (2020), https://www.popularmechanics.com/science/math/a32957375/mathematicians-boycott-predictive-policing/ (last visited Nov 9, 2021).
 Vidushi Marda & Shivangi Narayan, Data in New Delhi’s predictive policing system, in Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency 317–324 (2020), https://dl.acm.org/doi/10.1145/3351095.3372865 (last visited Nov 9, 2021).
 Predictive Policing: What is it, How it works, and its Legal Implications — The Centre for Internet and Society, supra note 62.
 Lt. Col. Nitisha v. Union of India, 2021 S.C.C. OnLine SC 261 (India).
 Madhu v. Northern Railways, 247 (2018) D.L.T. 198 (India).
 European Liberties Platform, 4 Benefits And 4 Drawbacks Of Predictive Policing, Liberties.eu , https://www.liberties.eu/en/stories/predictive-policing/43679 (last visited Nov 9, 2021).
 J. K. S. Puttaswamy (Retd.) & Anr. V. Union of India & Ors., (2017) 10 S.C.C. 1 (India).
 Jessica Saunders, Pitfalls of Predictive Policing (2016), https://www.rand.org/blog/2016/10/pitfalls-of-predictive-policing.html (last visited Nov 10, 2021).
 Predictive policing in India: Deterring crime or discriminating minorities?, , LSE Human Rights (2021), https://blogs.lse.ac.uk/humanrights/2021/04/16/predictive-policing-in-india-deterring-crime-or-discriminating-minorities/ (last visited Nov 7, 2021).
 Twitter, LAPD pioneered predicting crime with data. Many police don’t think it works, Los Angeles Times (2019), https://www.latimes.com/local/lanow/la-me-lapd-precision-policing-data-20190703-story.html (last visited Nov 10, 2021).
 Germany, , Automating Society Report 2020 , https://automatingsociety.algorithmwatch.org/report2020/germany (last visited Nov 10, 2021).
 How Predictive Policing Can Criminalize Kids’ Online Searches, , The Crime Report (2021), https://thecrimereport.org/2021/10/13/the-criminalization-of-curiosity/ (last visited Nov 9, 2021).
 Geetha N, ikotkur• March 25 & 2019 15 Minutes, Predictive Policing Can Obviate Cybercrime, https://www.bankinfosecurity.asia/interviews/predictive-policing-obviate-cybercrime-i-4278 (last visited Nov 10, 2021).