我正在尝试如下实现逻辑回归:
但是我无法得到好的预测,因为我的类输出 1 在我的数据中的代表性不足。因此,我试图将 SMOTE 算法应用于我的训练集以获得更好的结果。但是我收到消息错误:
Error in T[i, ] : subscript out of bounds
有我的代码:
set.seed(157)
split <- createDataPartition(df_statique$Y, p = .50,list = FALSE,times = 1)
trainSplit <- df_statique[ split,]
testSplit <- df_statique[-split,]
trainSplit <- SMOTE(Y ~ insolvency + efficiency + DebtToAssetsRatio + taille + CashAssetRatio + current + netWorth + REA, trainSplit, perc.over = 300, perc.under=100)
我的数据框 df_statique 有一部分:
index countryIsoCode insolvency efficiency CashAssetRatio DebtToAssetsRatio netWorth REA taille Y
41807 IT 0.00360 0.5193711 0.8686575 0.49446355 4387182 1.657145e-03 2 1
41808 IT 0.00050 1.5269309 1.6295765 0.36543122 30916838 6.601092e-03 3 0
41809 IT 0.00050 2.2635592 1.3427063 0.15809120 2200087 1.218576e-03 1 0
41810 IT 0.00280 1.3989753 0.9345793 0.69642554 2940473 3.852093e-04 2 0
41811 IT 0.00140 2.1440221 3.5781748 0.07951644 28418622 8.845920e-04 2 0
41812 IT 0.00040 1.0068491 1.7238305 0.47561418 22486133 2.703242e-04 2 0
41813 IT 0.00130 1.5569114 1.4459704 0.57632716 9769040 9.741611e-04 2 0
41814 IT 0.00510 5.0143711 0.1035034 0.71267895 3610152 2.391447e-03 2 0
41815 IT 0.00090 3.3280521 0.5160867 0.34998732 218965703 2.550272e-04 3 0
41816 IT 0.00040 1.7217051 2.2758391 0.29638050 29868519 1.136387e-04 3 0
41817 IT 0.00360 1.7261580 0.8490392 0.41231551 106020226 2.304773e-06 3 0
41818 IT 0.00040 1.3600893 1.6298656 0.57789518 55408765 4.841743e-04 3 1
41819 IT 0.00510 5.5565821 0.1376145 0.19679467 9491245 1.398124e-03 2 0
41820 IT 0.00131 3.8312347 1.1365521 0.73639696 8921497 4.701300e-06 3 0
41821 IT 0.00400 1.8218620 0.9113375 0.62646234 24134486 9.435248e-04 3 0
41822 IT 0.00100 1.8215702 1.0690901 0.82764828 777547 6.335832e-03 2 0
41823 IT 0.00090 1.8153513 0.9320536 0.80258849 2437903 6.035954e-04 2 0
41824 IT 0.00050 2.1300765 1.7388457 0.31394248 27009000 3.507500e-04 3 0
41825 IT 0.00100 1.8697385 1.4438289 0.56198890 35917 5.765082e-03 1 0
41826 IT 0.00230 6.5298138 1.1726536 0.56654516 2675415 1.038839e-02 2 0
41827 IT 0.00220 9.8201528 0.4794298 0.63618554 488924 1.336866e-05 2 0
最后,我的输出 Y 是一个虚拟变量,表示在 1 年内是否默认