sg平滑 matlab语言

clc;clear;close all
disp（‘原始数据，不同预处理方法进行处理的RMSECVRMSEP训练集及预测集的R‘）
load UV
cal    = calset（:1:600）;
caltar = calsettar;
for columm  = 1:4;
    dataname   = ynames{columm};
    disp（dataname）;      
   %----------------------------因子数的确定--------------------------------
   switch columm
    case 1
         maxrank = 8;
        
    case 2
         maxrank = 6;
         
    case 3
         maxrank = 7;
        
    case 4
         maxrank = 6;    
    end 
    [mn]                     = size（cal）;
    %-----------------KS对数据划分，训练集2/3，预测集1/3---------------------
    [modeltest]                    = kenstone（calfloor（2/3*m））;
    x_train                         = cal（model:）;
    x_pred                          = cal（test:）;
    y_train                         = caltar（modelcolumm）;
    y_pred                          = caltar（testcolumm）;
    [m_trainn_train]               = size（x_train）;
    %-------------------------------PLS------------------------------------
    disp（‘偏最小二乘-PLS‘）
    b                               = simpls（x_trainy_trainmaxrank）;
    c                               = x_pred*b（:maxrank）;
    [c_trainvalpresses]            = loocv（x_trainy_trainmaxrank）;
    disp（‘RMSECVRMSEP训练集及预测集的R‘）;
    rmsecv_PLS                      = sqrt（valpresses./m_train）;
    rmsep_PLS                       = rms（c-y_pred）;
    r_train                         = corrcoef（c_trainy_train）;
    r_train                         = r_train（12）;
    r_PLS                           = corrcoef（cy_pred）;
    r_PLS                           = r_PLS（12）;
    disp（[rmsecv_PLSrmsep_PLSr_trainr_PLS]）;
    disp（‘回收率最大最小值‘）;
    recovery_PLS                    = 100*c./y_pred;
    max_recovery_PLS（columm）        = max（recovery_PLS）;
    min_recovery_PLS（columm）        = min（recovery_PLS）;
    disp（[max_recovery_PLS（columm）min_recovery_PLS（columm）]）;
    clear b c c_train valpresses r_train 
    %-----------------------------