Fig. 1

Predictor feature selection using the LASSO logistic regression model

A Te tuning parameter(λ) was determined in the LASSO model by using a tenfold crossvalidation and a minimum criterion. In Figure A, the lower abscissa is log (lambda), the upper abscissa is the number of non-zero coefficients in the model, and the ordinate is the coefficient of the predictor. The different colored curves represent the trajectories of 11 different predictor coefficients

B The LASSO coefcient profle plot was generated against the log (lambda) sequence. In Figure B, the bottom horizontal coordinate is the logarithm of the penalty coefficient log (λ), and the top horizontal coordinate represents the number of predictors left in the equation for different λ. The vertical coordinate is the Mean-Squared error. The dashed line on the left is λmin, representing λ with the smallest deviation. The number of predictors is 5, when the model has the highest fitting effect