Analysis of the effect of life habits in the spreading process of Covid-19 with artificial intelligence

Pandemics are events that greatly affect the social and economic conditions of societies, and identifying the factors that affect their spread can provide effective clues and important information for coping with them. In this thesis, the behavioral characteristics and habits of infected and non-infected individuals during the COVID-19 pandemic in Turkey were analyzed using classification and association rule methodology. The classification results show that the status of being infected with the virus (the "Yes" class) and not being infected (the "No" class) can be classified using the Random Committe algorithm. In addition, it was seen that the classification performance obtained by using 42 attributes could be further increased and reduced to 14 attributes. Furthermore, it was found that the status of being infected can be predicted by the RRF (Regularized Random Forest) algorithm with an accuracy rate of 81% and a kappa value of 0.3663 using the reduced attributes, with an increased performance rate. In the second part of the study, association rule analyses were carried out to discover the relationships between the attributes, and 21 rules were found for the "Yes" class. A large number of rules were obtained for the "No" class for the status of being infected with COVID-19, and by increasing the support count value, the number of rules was reduced, resulting in a total of 2805 rules. The findings show that individuals living with one person in the household, having close contact 1-3 times a day, and having close contact outside the home for 4-6 hours are strong associations for the "Yes" class. For the "No" class, it was found that individuals who avoid traveling unless necessary, do not change their body weight since the beginning of the epidemic, work from home instead of going to work, and sometimes avoid small social groups in open areas show strong associations with the "No" class. In conclusion, this thesis has shown that the habits of life have an impact on the transmission and spread of epidemics and that these can be modeled using artificial intelligence techniques.