Teaching Documents by letezgi hagos
Abstract
The aim of this research to extract meaningful crime trends regarding offences against c... more Abstract
The aim of this research to extract meaningful crime trends regarding offences against children from the data in existing police records with the help of data mining techniques. We know children are exposed to different offences but we do not know which children are exposed to what type of offence (crime category). The output from this research helps to identify which children are exposed to which crime categories. Currently the police officers try to understand the relation between any two attributes but they do not know the relation among more than two attributes and the relationship between other variables and a class variable. This is why this can be achieved by using data mining techniques in an efficient and accurate manner than those achieved by trained personnel and traditional simple statistics to analyze crime data. The researcher used the six-phased CRISP-DM for data mining process and each of the steps in this model starting from business understanding up to evaluation and deployment phases are performed step wise and iteratively when needed. Even though all the phases are equally important the data pre-processing part has got due emphasis since police records are inconsistent and frequently incomplete making task of formal analysis inaccurate and time consuming. These analytical processes would benefit from using data mining techniques in a structured approach. Both unsupervised and supervised learning are used within the structured methodology to mine the police data. This research will serve as a reference material for researchers, crime investigators, planners and NGOs that work on prevention and control of offences against children. Based on this, it can also help to implement different crime preventive programs like through awareness creation programs.
The research demonstrates that data mining techniques can be successfully used in proactive policing to prevent crimes. This is more applicable for high volume crimes such as theft, violence and sexual assaults that have been committed most commonly. These crimes can often be segmented and classified and the generated models can be used to
predict potential victims of a specified crime category through predictive models as well as to attribute the profile of victims with the help of descriptive techniques.
Some of the rules in association rule are not interesting due to few values that are unable to generate patterns. From all the crime categories in the crime records sexual assault has the highest number and best rules are generated related with sexual assault. Almost all interesting rules generated in association rule are included in the rules generated by the classification model. Generally, promising results are registered with encourage further researches in the area.
Papers by letezgi hagos
Uploads
Teaching Documents by letezgi hagos
The aim of this research to extract meaningful crime trends regarding offences against children from the data in existing police records with the help of data mining techniques. We know children are exposed to different offences but we do not know which children are exposed to what type of offence (crime category). The output from this research helps to identify which children are exposed to which crime categories. Currently the police officers try to understand the relation between any two attributes but they do not know the relation among more than two attributes and the relationship between other variables and a class variable. This is why this can be achieved by using data mining techniques in an efficient and accurate manner than those achieved by trained personnel and traditional simple statistics to analyze crime data. The researcher used the six-phased CRISP-DM for data mining process and each of the steps in this model starting from business understanding up to evaluation and deployment phases are performed step wise and iteratively when needed. Even though all the phases are equally important the data pre-processing part has got due emphasis since police records are inconsistent and frequently incomplete making task of formal analysis inaccurate and time consuming. These analytical processes would benefit from using data mining techniques in a structured approach. Both unsupervised and supervised learning are used within the structured methodology to mine the police data. This research will serve as a reference material for researchers, crime investigators, planners and NGOs that work on prevention and control of offences against children. Based on this, it can also help to implement different crime preventive programs like through awareness creation programs.
The research demonstrates that data mining techniques can be successfully used in proactive policing to prevent crimes. This is more applicable for high volume crimes such as theft, violence and sexual assaults that have been committed most commonly. These crimes can often be segmented and classified and the generated models can be used to
predict potential victims of a specified crime category through predictive models as well as to attribute the profile of victims with the help of descriptive techniques.
Some of the rules in association rule are not interesting due to few values that are unable to generate patterns. From all the crime categories in the crime records sexual assault has the highest number and best rules are generated related with sexual assault. Almost all interesting rules generated in association rule are included in the rules generated by the classification model. Generally, promising results are registered with encourage further researches in the area.
Papers by letezgi hagos
The aim of this research to extract meaningful crime trends regarding offences against children from the data in existing police records with the help of data mining techniques. We know children are exposed to different offences but we do not know which children are exposed to what type of offence (crime category). The output from this research helps to identify which children are exposed to which crime categories. Currently the police officers try to understand the relation between any two attributes but they do not know the relation among more than two attributes and the relationship between other variables and a class variable. This is why this can be achieved by using data mining techniques in an efficient and accurate manner than those achieved by trained personnel and traditional simple statistics to analyze crime data. The researcher used the six-phased CRISP-DM for data mining process and each of the steps in this model starting from business understanding up to evaluation and deployment phases are performed step wise and iteratively when needed. Even though all the phases are equally important the data pre-processing part has got due emphasis since police records are inconsistent and frequently incomplete making task of formal analysis inaccurate and time consuming. These analytical processes would benefit from using data mining techniques in a structured approach. Both unsupervised and supervised learning are used within the structured methodology to mine the police data. This research will serve as a reference material for researchers, crime investigators, planners and NGOs that work on prevention and control of offences against children. Based on this, it can also help to implement different crime preventive programs like through awareness creation programs.
The research demonstrates that data mining techniques can be successfully used in proactive policing to prevent crimes. This is more applicable for high volume crimes such as theft, violence and sexual assaults that have been committed most commonly. These crimes can often be segmented and classified and the generated models can be used to
predict potential victims of a specified crime category through predictive models as well as to attribute the profile of victims with the help of descriptive techniques.
Some of the rules in association rule are not interesting due to few values that are unable to generate patterns. From all the crime categories in the crime records sexual assault has the highest number and best rules are generated related with sexual assault. Almost all interesting rules generated in association rule are included in the rules generated by the classification model. Generally, promising results are registered with encourage further researches in the area.