| Abstract |
A novel variable selection technique termed as CARS-Logistic is proposed by fusing
competitive adaptive re-weighted sampling(CARS) and logistic regression. In the current
study, the modulus of regression coefficients of the logistic model is utilized for assessing
the significance of variables. The capability of CARS-Logistic is assessed using two data
sets: real-life and SIMUL datasets. The outcomes disclose that the proposed method is efficient
than classical variable selection methods. It can select more significant variables with
the least AIC and BIC values together with greater values of three Pseudo R-squared(s).
The study also points out the factors leading to perinatal mortality in Pakistan. The identified
hazards communicate social, cultural, financial, and health-related characteristics of
mothers.
xv |
