International Research Symposium on Pure and Applied Sciences (IRSPAS)

Permanent URI for this communityhttp://repository.kln.ac.lk/handle/123456789/15650

Browse

Subject: search.filters.subject.Multinomial logistic regression

Search Results

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Factors associated with severity of road accidents in Gampaha police division in Sri Lanka
    (Research Symposium on Pure and Applied Sciences, 2018 Faculty of Science, University of Kelaniya, Sri Lanka, 2018) Madhubhashini, M. G. A. K.; Attanayake, A. M. C. H.; Sooriyaarachchi, D. J. C.
    Road accidents are main social problem in Sri Lanka, which causes immense damages and injuries unintentionally and unexpectedly. There are numerous factors, which contribute road accidents such as time of the day, road surface, weather, light condition, location, traffic density, traffic control availability, vehicle type, vehicle age, driver’s gender, driver’s age, license availability, usage of alcohol and etc. This study aims to identify the factors that mainly contribute to road accident severity in Gampaha Police Division through formulating an effective model. Altogether, 1375 data were collected from 2015 to 2017 from Police report of the Gampaha Police Division and the above mentioned 13 factors were considered. In this research 25% of the data were used for the validate the model and remain 75% of the data were used for build the model. Severity of accidents were categorized as fatal, grievous, damage only and non-grievous accidents. Chi-square test of independence has detected that road surface, light condition, location, traffic density, traffic control availability, vehicle type, driver’s age and usage of alcohol are the significant factors. The Location variable removed due to multicollinearity and remain significant factors used for Multinomial logistic model to model the severity of road accidents. The area under the ROC value was 59.2% for model building data and the area under the ROC value was 58.4% for model validation data. That means the developed model more accurately predicts the severity of accidents than the prediction in baseline model. Based on the results, it is discovered that traffic control availability increases the effect on the probability of a fatal accidents and fail of alcohol test, road surface and vehicle type decreases the effect on the probability of a fatal accidents. Moreover, traffic control availability and traffic density increase the effect on the probability of grievous category and alcohol test, vehicle type have negative impact on the probability of damage only accidents. Traffic density have positive impact on damage only accidents. The analysis shows different factors contribute a significant impact on the severity of road accidents. The result of this research is useful for police to understand factors affect on severity of road accidents and decrease the number of road accidents in future.
  • Thumbnail Image
    Item
    Factors that affect labour induction and its successfulness of pregnancies in Sri Lanka.
    (International Research Symposium on Pure and Applied Sciences, 2017 Faculty of Science, University of Kelaniya, Sri Lanka., 2017) Gunawardana, J. R. N. A.; Perera, S. S. N.
    Induction of Labour (IOL) is an important practice that is carried out commonly in modern day obstetrics. In medium to large health care facilities, it is estimated that approximately 35.5% of all deliveries involve IOL in Sri Lanka. The main objective of this study, was to identify the factors that affect IOL and to assess the association between IOL and each pregnancy outcome. In this study, we consider data of all women who were admitted to selected health care facilities for delivery in 3 randomly selected provinces in Sri Lanka, for the period from July to October 2011. Multinomial Logistic Regression model (MLR) and Fuzzy Expert System were used to identify the factors that affect IOL. MLR model predicts for spontaneous labour group and induced labour group, with reference to no labour (caesarean section/C-section) category. Obtained score under Fuzzy Expert System was appropriate to distinguish whether an individual should go through IOL or not. It also can be used to identify whether a new born would survive after seven days of life. The MLR model predicts for IOL with a classification rate of 65.5% and the Fuzzy Expert System predicts for IOL with an accuracy of 55.10%. Results indicated that IOL was related to maternal age, number of previous caesarian sections, number of previous births, estimate gestational age, number of previous pregnancies, PreEclampsia, Placenta Preavia, Abruption Placenta, total number of neonates delivered, birth weight and Maternal Severity Index (MSI). Fuzzy Expert System also states that, if the score is between 0.8570 and 0.8854, then the patient will belong to induced group and the baby would be alive after seven days of birth. This study concludes that, MLR and Fuzzy models can be used to deal with decision making procedures related to IOL.