- Setting the seed and partitioning the data. The proportion of observations in the training data set should be 70%. The remaining 30% of observations should be in the test data set.
#Setting the seed
set.seed(71923)
#Train Test Split
train_insts = sample(nrow(carDF), .7*nrow(carDF))
carDFTrain <- carDF[train_insts,]
carDFTest <- carDF[-train_insts,]- Boxplots of VehOdo and VehicleAge (broken up by values of IsBadBuy).
#Plotting the box plots
ggplot(carDFTrain, aes( x = IsBadBuy, y = VehOdo)) + geom_boxplot()
ggplot(carDFTrain, aes(x = IsBadBuy, y = VehicleAge)) + geom_boxplot()
We can see that bad buys have a higher vehicle age.
- Report a two-way table for count of IsBadBuy broken up by Make (i.e. what are the numbers of each make of cars that are good vs. bad buys?)
#Two-way table
table(carDFTrain$Make, carDFTrain$IsBadBuy)##
## 0 1
## ACURA 2 4
## BUICK 31 38
## CADILLAC 0 2
## CHEVROLET 826 623
## CHRYSLER 425 430
## DODGE 637 540
## FORD 550 690
## GMC 29 34
## HONDA 31 27
## HYUNDAI 75 77
## INFINITI 2 5
## ISUZU 7 4
## JEEP 75 94
## KIA 152 115
## LEXUS 0 4
## LINCOLN 5 8
## MAZDA 52 64
## MERCURY 46 65
## MINI 3 4
## MITSUBISHI 54 50
## NISSAN 90 136
## OLDSMOBILE 7 25
## PLYMOUTH 0 1
## PONTIAC 226 187
## SATURN 98 116
## SCION 10 6
## SUBARU 1 2
## SUZUKI 64 77
## TOYOTA 50 51
## VOLKSWAGEN 6 7
## VOLVO 3 0
For Ford vehicles bad buys are really high than good buys. But we are not really controlling for other variables hence cannot make a concrete argument here.
- Training a logistic regression model of IsBadBuy on the demographic variables using the training data set.
#Logistic Regression
RegressionPart4 <- glm(IsBadBuy ~ Auction + VehicleAge + Make +Color +
WheelType + VehOdo + Size + MMRAcquisitionAuctionAveragePrice +
MMRAcquisitionRetailAveragePrice, data = carDFTrain, family = "binomial")
summary(RegressionPart4)##
## Call:
## glm(formula = IsBadBuy ~ Auction + VehicleAge + Make + Color +
## WheelType + VehOdo + Size + MMRAcquisitionAuctionAveragePrice +
## MMRAcquisitionRetailAveragePrice, family = "binomial", data = carDFTrain)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.9194 -0.9712 -0.5640 1.0977 2.1419
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.182e+00 1.279e+00 -1.707 0.08785 .
## AuctionMANHEIM 2.017e-01 7.169e-02 2.813 0.00491 **
## AuctionOTHER 1.285e-02 8.601e-02 0.149 0.88128
## VehicleAge 2.716e-01 2.745e-02 9.896 < 2e-16 ***
## MakeBUICK 1.540e-01 9.669e-01 0.159 0.87348
## MakeCADILLAC 1.324e+01 3.013e+02 0.044 0.96495
## MakeCHEVROLET -1.071e-01 9.334e-01 -0.115 0.90864
## MakeCHRYSLER 3.758e-01 9.319e-01 0.403 0.68678
## MakeDODGE 1.293e-01 9.304e-01 0.139 0.88951
## MakeFORD 2.647e-01 9.326e-01 0.284 0.77651
## MakeGMC 9.698e-03 9.704e-01 0.010 0.99203
## MakeHONDA -3.037e-01 9.758e-01 -0.311 0.75562
## MakeHYUNDAI 7.513e-02 9.482e-01 0.079 0.93684
## MakeINFINITI 1.291e+00 1.337e+00 0.966 0.33418
## MakeISUZU -8.300e-01 1.200e+00 -0.692 0.48902
## MakeJEEP 2.201e-01 9.512e-01 0.231 0.81703
## MakeKIA 1.342e-01 9.429e-01 0.142 0.88684
## MakeLEXUS 1.432e+01 2.455e+02 0.058 0.95350
## MakeLINCOLN 1.366e-01 1.103e+00 0.124 0.90145
## MakeMAZDA 1.787e-01 9.527e-01 0.188 0.85118
## MakeMERCURY 1.908e-01 9.524e-01 0.200 0.84122
## MakeMINI 3.541e-01 1.246e+00 0.284 0.77624
## MakeMITSUBISHI -3.189e-01 9.570e-01 -0.333 0.73897
## MakeNISSAN 3.216e-01 9.415e-01 0.342 0.73266
## MakeOLDSMOBILE 7.068e-01 1.033e+00 0.684 0.49388
## MakePLYMOUTH 1.225e+01 5.354e+02 0.023 0.98174
## MakePONTIAC -1.584e-02 9.376e-01 -0.017 0.98652
## MakeSATURN 1.484e-01 9.425e-01 0.157 0.87486
## MakeSCION 6.910e-02 1.089e+00 0.063 0.94942
## MakeSUBARU 7.433e-01 1.554e+00 0.478 0.63231
## MakeSUZUKI 6.165e-01 9.526e-01 0.647 0.51755
## MakeTOYOTA 7.133e-02 9.574e-01 0.075 0.94061
## MakeVOLKSWAGEN 2.376e-01 1.107e+00 0.215 0.83000
## MakeVOLVO -1.299e+01 3.026e+02 -0.043 0.96577
## ColorBEIGE -8.920e-01 7.975e-01 -1.119 0.26333
## ColorBLACK -7.628e-01 7.818e-01 -0.976 0.32922
## ColorBLUE -8.668e-01 7.807e-01 -1.110 0.26687
## ColorBROWN -7.236e-01 8.349e-01 -0.867 0.38614
## ColorGOLD -5.766e-01 7.838e-01 -0.736 0.46191
## ColorGREEN -9.753e-01 7.877e-01 -1.238 0.21564
## ColorGREY -8.267e-01 7.817e-01 -1.058 0.29028
## ColorMAROON -5.030e-01 7.938e-01 -0.634 0.52625
## ColorNULL -4.090e+00 1.677e+00 -2.439 0.01472 *
## ColorORANGE -6.950e-01 8.844e-01 -0.786 0.43193
## ColorOTHER -2.059e+00 9.295e-01 -2.215 0.02673 *
## ColorPURPLE -1.264e-01 8.987e-01 -0.141 0.88811
## ColorRED -7.529e-01 7.825e-01 -0.962 0.33591
## ColorSILVER -6.722e-01 7.793e-01 -0.863 0.38835
## ColorWHITE -6.990e-01 7.801e-01 -0.896 0.37017
## ColorYELLOW -1.169e+00 8.999e-01 -1.299 0.19388
## WheelTypeCovers -1.015e-01 6.463e-02 -1.570 0.11636
## WheelTypeNULL 3.444e+00 1.612e-01 21.368 < 2e-16 ***
## WheelTypeSpecial -9.917e-03 2.502e-01 -0.040 0.96838
## VehOdo 1.186e-05 2.087e-06 5.684 1.32e-08 ***
## Size'LARGE TRUCK' 8.249e-02 2.312e-01 0.357 0.72129
## Size'MEDIUM SUV' 2.746e-01 2.112e-01 1.300 0.19352
## Size'SMALL SUV' 1.266e-01 2.591e-01 0.489 0.62503
## Size'SMALL TRUCK' 5.029e-01 3.035e-01 1.657 0.09747 .
## SizeCOMPACT 6.546e-01 2.348e-01 2.788 0.00530 **
## SizeCROSSOVER -2.993e-01 2.696e-01 -1.110 0.26683
## SizeLARGE 1.170e-01 2.252e-01 0.519 0.60346
## SizeMEDIUM 2.973e-01 2.203e-01 1.349 0.17724
## SizeSPECIALTY 3.650e-01 2.772e-01 1.316 0.18802
## SizeSPORTS 1.833e-01 3.049e-01 0.601 0.54773
## SizeVAN 1.944e-01 2.307e-01 0.843 0.39940
## MMRAcquisitionAuctionAveragePrice 3.842e-06 3.474e-05 0.111 0.91194
## MMRAcquisitionRetailAveragePrice 1.495e-06 2.155e-05 0.069 0.94469
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 9763.0 on 7042 degrees of freedom
## Residual deviance: 8124.3 on 6976 degrees of freedom
## AIC: 8258.3
##
## Number of Fisher Scoring iterations: 12
ConfusionMatrixTwo <- confusion_matrix(preds = predict(RegressionPart4, carDFTest, type = "response"), actuals = carDFTest$IsBadBuy, cutoff = .5)
ConfusionMatrixTwo## classifications
## actuals 0 1
## 0 1175 326
## 1 643 875