PERBANDINGAN VARIABEL DOMINAN FAKTOR RISIKO KEJADIAN BERAT BADAN LAHIR RENDAH ANTARA HASIL ANALISIS REGRESI LOGISTIK DAN POHON KLASIFIKASI

• Main Author: PIPIT FESTI W, 090810147
• Format: Theses and Dissertations NonPeerReviewed
• Language: English; Indonesian
• Published: 2010
• Subjects: QA276-280 Mathematical Analysis; QM Human anatomy; RB Pathology
• Online Access: http://repository.unair.ac.id/37944/1/gdlhub-gdl-s2-2011-pipitfesti-19145-tkm051-k.pdf; http://repository.unair.ac.id/37944/2/gdlhub-gdl-s2-2011-pipitfesti-15934-tkm0511.pdf; http://repository.unair.ac.id/37944/; http://lib.unair.ac.id
• Institution: Universitas Airlangga

In health research that studies the influence of several determinants of an event used the regression method. One type of regression is a logistic regression model that is a mathematical approach that can be used to describe the relationship between the dichotomous dependent variable or polychotomus with dichotomous independent variables, polythomous and continuous. That Method has limitations on the processing of health data that can be addressed by the method of classification tree. This research is a statistical study comparing the dominant variable risk factors event of LBW (low birth body weight) between the results of logistic regression analysis and classification tree. This research was applied on secondary data of Birth Weight at Health Department of Sumenep City and the cohort report of pregnant women at five health centers in Sumenep district. Dependent variable is low birth weight and the independent variable is maternal age, maternal education, maternal employment status, numbers of children, birth spacing, maternal hemoglobin, mothers LILA size (Mother Upper Arm Circumference), increase in maternal weight and maternal height. The amount of data in this study was 337 data. Logistic regression analysis with #945; = 0.05 independent variables that affect the LBW were maternal education, maternal hemoglobin, LILA mother and Body Weight. Results of classification tree analysis on optimal tree were increase of maternal weight gain and maternal education. Analysis on maximum tree obtained the dominant variables such as increase maternal weight gain, maternal education, mothers LILA size, maternal Hemoglobin, numbers of children and total maternal weight. Result of classification accuracy showed that the logistics regression has a higher accuracy of the classification than classification tree method. The result of logistic regression classification accuracy is 80.7% higher than the optimum classification tree, 71.5% and 71.9% on maximum classification tree.