@phdthesis{digilib69944,
           month = {January},
           title = {OPTIMISASI ANALISIS REGRESI LASSO DENGAN ALGORITMA LARS MENGGUNAKAN BAYESIAN INFORMATION CRITERION (BIC) STUDI KASUS: PENDUGAAN DATA INDEKS KETAHANAN PANGAN TAHUN 2021},
          school = {UIN SUNAN KALIJAGA YOGYAKARTA},
          author = {NIM.: 20106010042 Zulaeka Ritasari},
            year = {2025},
            note = {Sri Utami Zuliana, S.Si.,M.Sc.,Ph.D.},
        keywords = {Multikolinearitas, LASSO, LARS, BIC, Indeks Ketahanan Pangan},
             url = {https://digilib.uin-suka.ac.id/id/eprint/69944/},
        abstract = {Correlation between predictor variables in regression analysis can be a problem
or violation of linear regression assumptions that called multicollinearity. Multicollinearity
causes the variance to become large, so that the estimation of regression
parameters becomes inefficient. Multicollinearity can be detection by considering
the variance inflation factor (VIF) value and its correlation coefficient. This
problem can be overcome by using Least Absolute Shrinkage and Selection Operator
(LASSO) Regression. LASSO will shrink the regression coefficient close to
zero or exactly zero, the calculation of which is done with the help of the Least Absolute
Regression (LARS) algorithm. In the case study on the Food Security Index
data, predictor variables that have high correlation will be gradually entered into the
model. The first variable to enter is X4 then continued by X9, X2, X1, X8, X6, X11,
X7, X5, X3 up to X10. The results of the coefficient estimation at each stage will
produce several regression models. These stages produce several regression models
which are then tested by selecting the best Bayesian Information Criterion (BIC)
model. The minimum BIC value (0.342) indicates an optimal lambda of 0.059, so
the best model is obtained at the thirty-fifth stage. The results of the study show that
seven of the eleven predictor variables have an influence on the food security index
in 2021.}
}