Predicting Bankruptcy of a company
Kunal Priyadarshi
February 16, 2019
Introduction
Problem
Predict whether a company will default based on its financial performance using logistic, stepwise - forward and backward regression.
Approach
We will follow these steps to classify companies for default cases:
- Random sample a training data set that contains 75% of original data points
- Exploratory data analysis
- Univariate
- Bivariate
- Find a best model for Credit Scoring data using logistic regression with AIC and BIC
- Model Performance
- Draw ROC curve
- Report the AUC
- Present the misclassification rate table of the final model
- Test the out-of-sample performance: Using final logistic linear model built from 75% of original data, we will test the model on the remaining 25% data
- Report out-of-sample AUC and asymmetric misclassification rate
Exploratory Data Analysis
Data Preparation and Summary
Data contains 5436 observations and 13 variables. There are no missing values in the data. For the purpose of this analysis we are ignoring the FYEAR and CUSIP columns as they are not relevant for our analysis.
library(ggplot2)
library(dplyr)
library(tidyr)
library(corrplot)
#### Input and summary of data ###
Bankruptcy <- read.csv("Bankruptcy.csv")
Bank_clean <- Bankruptcy[, -c(1, 3)]Sample data: first six rows
head(Bankruptcy)## FYEAR DLRSN CUSIP R1 R2 R3 R4
## 1 1999 0 00036020 0.3071395 0.8870057 1.6476808 -0.19915760
## 2 1999 0 00036110 0.7607365 0.5924934 0.4530028 -0.36989014
## 3 1999 0 00037520 -0.5135961 0.3376148 0.2990145 -0.02907996
## 4 1994 1 00078110 -0.4661293 0.3707467 0.4960667 -0.37342897
## 5 1999 0 00079X10 2.0234223 0.2148764 0.1825948 6.69536040
## 6 1999 0 00086T10 0.9074985 0.3868797 0.4778914 -0.34715872
## R5 R6 R7 R8 R9 R10
## 1 1.0929640 -0.31328867 -0.19679332 1.2067628 0.2824709 0.15889625
## 2 0.1861539 0.03961907 0.32749732 0.4284181 1.1069652 0.79344276
## 3 -0.4326047 0.82999324 -0.70778613 0.4761533 2.1791755 2.48458450
## 4 -0.2674240 0.97779888 -0.61097507 0.4568098 0.1519511 0.04778851
## 5 -1.1483381 -1.50588927 2.87647687 0.2873749 -0.9864421 0.79107673
## 6 1.4079893 -0.48390230 0.07025944 0.5278110 0.5024659 -0.16464802
Structure of the data
str(Bankruptcy)## 'data.frame': 5436 obs. of 13 variables:
## $ FYEAR: int 1999 1999 1999 1994 1999 1999 1999 1999 1987 1999 ...
## $ DLRSN: int 0 0 0 1 0 0 0 0 1 0 ...
## $ CUSIP: Factor w/ 5436 levels "00036020","00036110",..: 1 2 3 4 5 6 7 8 9 10 ...
## $ R1 : num 0.307 0.761 -0.514 -0.466 2.023 ...
## $ R2 : num 0.887 0.592 0.338 0.371 0.215 ...
## $ R3 : num 1.648 0.453 0.299 0.496 0.183 ...
## $ R4 : num -0.1992 -0.3699 -0.0291 -0.3734 6.6954 ...
## $ R5 : num 1.093 0.186 -0.433 -0.267 -1.148 ...
## $ R6 : num -0.3133 0.0396 0.83 0.9778 -1.5059 ...
## $ R7 : num -0.197 0.327 -0.708 -0.611 2.876 ...
## $ R8 : num 1.207 0.428 0.476 0.457 0.287 ...
## $ R9 : num 0.282 1.107 2.179 0.152 -0.986 ...
## $ R10 : num 0.1589 0.7934 2.4846 0.0478 0.7911 ...
Distribution of bankruptcy - 0’s and 1’s in the data (1 means bankrupt)
table(Bankruptcy$DLRSN)##
## 0 1
## 4660 776Dimension of data
dim(Bankruptcy)## [1] 5436 13Distribution of 0’s and 1’s in the data over the years
table(Bankruptcy$FYEAR, Bankruptcy$DLRSN)##
## 0 1
## 1980 0 24
## 1981 0 36
## 1982 0 20
## 1983 0 41
## 1984 0 55
## 1985 0 48
## 1986 0 47
## 1987 0 44
## 1988 0 49
## 1989 0 73
## 1990 0 77
## 1991 0 47
## 1992 0 35
## 1993 0 33
## 1994 0 31
## 1995 0 40
## 1996 0 51
## 1997 0 24
## 1998 0 1
## 1999 4660 0
Univariate Analysis
Long <- Bank_clean %>% gather()
ggplot(Long, aes(value)) + facet_wrap(~ key) + geom_histogram()
- We see that R4, R5 follow a right skewed distribution.
- R1, R2 follow a left skewed distribution.
- R10, R3, R4 R8, R7 and R9 follow an almost normal distribution with a few extreme values. We make a boxplot to verify the outliers and extreme values.
Long <- Bank_clean %>% gather()
ggplot(Long, aes(x = key, y = value)) + geom_boxplot()
- As also seen from the boxplot above, we observe outliers in the variable – R1, R4, R5, R6, R7.
Bivariate Analysis
Correlation <- cor(Bank_clean)
corrplot(Correlation)
Response Variable DLRSN is highly correlated with variable R10 followed by R8, R1 and R3.
Model Selection
We will now employ different methods to find the best model.
Logistic Regression:Summary
### Sampling ###
set.seed(12957846)
index <- sample(nrow(Bank_clean), 0.75*nrow(Bank_clean))
Bankruptcy_train <- Bank_clean[index,]
Bankruptcy_test <- Bank_clean[-index,]
### Model Building ###
# Logistic #
model_logistic <- glm(DLRSN ~., data = Bankruptcy_train, family = "binomial")
summary(model_logistic)##
## Call:
## glm(formula = DLRSN ~ ., family = "binomial", data = Bankruptcy_train)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.2267 -0.4653 -0.2484 -0.1154 3.3268
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.47076 0.07762 -31.833 < 2e-16 ***
## R1 0.23062 0.07690 2.999 0.002709 **
## R2 0.54190 0.07988 6.784 1.17e-11 ***
## R3 -0.35785 0.10055 -3.559 0.000372 ***
## R4 -0.09394 0.10360 -0.907 0.364528
## R5 0.02784 0.05359 0.520 0.603349
## R6 0.23985 0.05685 4.219 2.46e-05 ***
## R7 -0.49011 0.10819 -4.530 5.90e-06 ***
## R8 -0.35951 0.08607 -4.177 2.96e-05 ***
## R9 0.28565 0.10672 2.677 0.007433 **
## R10 -1.49359 0.09925 -15.048 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 3364.0 on 4076 degrees of freedom
## Residual deviance: 2308.7 on 4066 degrees of freedom
## AIC: 2330.7
##
## Number of Fisher Scoring iterations: 7Logistic model deviance = 2308.66
Logistic model AIC = 2330.66
Logistic model BIC = 2400.11
Stepwise - Forward, backward, both and backward with bic
# Variabel Selection #
NUll_model <- glm(DLRSN ~ 1, data = Bankruptcy_train, family = "binomial")
Full_model <- glm(DLRSN ~ ., data = Bankruptcy_train, family = "binomial")
Backward_selection <- step(Full_model, direction = "backward")## Start: AIC=2330.66
## DLRSN ~ R1 + R2 + R3 + R4 + R5 + R6 + R7 + R8 + R9 + R10
##
## Df Deviance AIC
## - R5 1 2308.9 2328.9
## - R4 1 2309.6 2329.6
## <none> 2308.7 2330.7
## - R9 1 2315.9 2335.9
## - R1 1 2317.8 2337.8
## - R3 1 2321.3 2341.3
## - R6 1 2326.6 2346.6
## - R8 1 2327.0 2347.0
## - R7 1 2331.4 2351.4
## - R2 1 2358.3 2378.3
## - R10 1 2559.7 2579.7
##
## Step: AIC=2328.93
## DLRSN ~ R1 + R2 + R3 + R4 + R6 + R7 + R8 + R9 + R10
##
## Df Deviance AIC
## - R4 1 2309.7 2327.7
## <none> 2308.9 2328.9
## - R1 1 2318.2 2336.2
## - R9 1 2320.5 2338.5
## - R3 1 2321.6 2339.6
## - R8 1 2327.3 2345.3
## - R6 1 2328.1 2346.1
## - R7 1 2331.9 2349.9
## - R2 1 2358.3 2376.3
## - R10 1 2642.8 2660.8
##
## Step: AIC=2327.69
## DLRSN ~ R1 + R2 + R3 + R6 + R7 + R8 + R9 + R10
##
## Df Deviance AIC
## <none> 2309.7 2327.7
## - R1 1 2319.4 2335.4
## - R3 1 2322.3 2338.3
## - R9 1 2325.3 2341.3
## - R8 1 2327.7 2343.7
## - R6 1 2330.2 2346.2
## - R7 1 2337.9 2353.9
## - R2 1 2361.2 2377.2
## - R10 1 2763.3 2779.3Forward_selection <- step(NUll_model, direction = "forward", scope =
list(lower = NUll_model, upper = Full_model))## Start: AIC=3365.99
## DLRSN ~ 1
##
## Df Deviance AIC
## + R10 1 2535.2 2539.2
## + R6 1 3053.5 3057.5
## + R8 1 3066.3 3070.3
## + R3 1 3097.0 3101.0
## + R1 1 3160.0 3164.0
## + R9 1 3193.3 3197.3
## + R7 1 3195.6 3199.6
## + R4 1 3196.9 3200.9
## + R2 1 3208.8 3212.8
## + R5 1 3251.3 3255.3
## <none> 3364.0 3366.0
##
## Step: AIC=2539.2
## DLRSN ~ R10
##
## Df Deviance AIC
## + R7 1 2470.4 2476.4
## + R6 1 2481.1 2487.1
## + R9 1 2487.6 2493.6
## + R8 1 2488.8 2494.8
## + R4 1 2489.3 2495.3
## + R3 1 2507.7 2513.7
## + R1 1 2508.4 2514.4
## + R5 1 2510.9 2516.9
## + R2 1 2533.0 2539.0
## <none> 2535.2 2539.2
##
## Step: AIC=2476.38
## DLRSN ~ R10 + R7
##
## Df Deviance AIC
## + R9 1 2437.8 2445.8
## + R8 1 2438.7 2446.7
## + R3 1 2450.2 2458.2
## + R4 1 2456.1 2464.1
## + R5 1 2456.1 2464.1
## + R6 1 2456.2 2464.2
## + R2 1 2460.0 2468.0
## <none> 2470.4 2476.4
## + R1 1 2469.5 2477.5
##
## Step: AIC=2445.77
## DLRSN ~ R10 + R7 + R9
##
## Df Deviance AIC
## + R8 1 2376.2 2386.2
## + R3 1 2379.0 2389.0
## + R6 1 2422.8 2432.8
## <none> 2437.8 2445.8
## + R4 1 2436.5 2446.5
## + R5 1 2436.6 2446.6
## + R2 1 2437.3 2447.3
## + R1 1 2437.4 2447.4
##
## Step: AIC=2386.21
## DLRSN ~ R10 + R7 + R9 + R8
##
## Df Deviance AIC
## + R2 1 2342.8 2354.8
## + R3 1 2370.4 2382.4
## + R4 1 2372.4 2384.4
## + R6 1 2373.5 2385.5
## <none> 2376.2 2386.2
## + R1 1 2375.1 2387.1
## + R5 1 2376.2 2388.2
##
## Step: AIC=2354.84
## DLRSN ~ R10 + R7 + R9 + R8 + R2
##
## Df Deviance AIC
## + R3 1 2330.3 2344.3
## + R6 1 2332.0 2346.0
## + R4 1 2340.8 2354.8
## <none> 2342.8 2354.8
## + R5 1 2341.9 2355.9
## + R1 1 2342.7 2356.7
##
## Step: AIC=2344.34
## DLRSN ~ R10 + R7 + R9 + R8 + R2 + R3
##
## Df Deviance AIC
## + R6 1 2319.4 2335.4
## + R4 1 2328.3 2344.3
## <none> 2330.3 2344.3
## + R5 1 2329.4 2345.4
## + R1 1 2330.2 2346.2
##
## Step: AIC=2335.44
## DLRSN ~ R10 + R7 + R9 + R8 + R2 + R3 + R6
##
## Df Deviance AIC
## + R1 1 2309.7 2327.7
## <none> 2319.4 2335.4
## + R4 1 2318.2 2336.2
## + R5 1 2319.2 2337.2
##
## Step: AIC=2327.69
## DLRSN ~ R10 + R7 + R9 + R8 + R2 + R3 + R6 + R1
##
## Df Deviance AIC
## <none> 2309.7 2327.7
## + R4 1 2308.9 2328.9
## + R5 1 2309.6 2329.6Both_selection <- step(NUll_model, direction = "both", scope =
list(lower = NUll_model, upper = Full_model))## Start: AIC=3365.99
## DLRSN ~ 1
##
## Df Deviance AIC
## + R10 1 2535.2 2539.2
## + R6 1 3053.5 3057.5
## + R8 1 3066.3 3070.3
## + R3 1 3097.0 3101.0
## + R1 1 3160.0 3164.0
## + R9 1 3193.3 3197.3
## + R7 1 3195.6 3199.6
## + R4 1 3196.9 3200.9
## + R2 1 3208.8 3212.8
## + R5 1 3251.3 3255.3
## <none> 3364.0 3366.0
##
## Step: AIC=2539.2
## DLRSN ~ R10
##
## Df Deviance AIC
## + R7 1 2470.4 2476.4
## + R6 1 2481.0 2487.0
## + R9 1 2487.6 2493.6
## + R8 1 2488.8 2494.8
## + R4 1 2489.3 2495.3
## + R3 1 2507.7 2513.7
## + R1 1 2508.4 2514.4
## + R5 1 2511.0 2517.0
## + R2 1 2533.0 2539.0
## <none> 2535.2 2539.2
## - R10 1 3364.0 3366.0
##
## Step: AIC=2476.38
## DLRSN ~ R10 + R7
##
## Df Deviance AIC
## + R9 1 2437.8 2445.8
## + R8 1 2438.7 2446.7
## + R3 1 2450.2 2458.2
## + R4 1 2456.1 2464.1
## + R5 1 2456.1 2464.1
## + R6 1 2456.2 2464.2
## + R2 1 2460.0 2468.0
## <none> 2470.4 2476.4
## + R1 1 2469.5 2477.5
## - R7 1 2535.2 2539.2
## - R10 1 3195.6 3199.6
##
## Step: AIC=2445.77
## DLRSN ~ R10 + R7 + R9
##
## Df Deviance AIC
## + R8 1 2376.2 2386.2
## + R3 1 2379.0 2389.0
## + R6 1 2422.8 2432.8
## <none> 2437.8 2445.8
## + R4 1 2436.5 2446.5
## + R5 1 2436.6 2446.6
## + R2 1 2437.3 2447.3
## + R1 1 2437.4 2447.4
## - R9 1 2470.4 2476.4
## - R7 1 2487.6 2493.6
## - R10 1 2993.5 2999.5
##
## Step: AIC=2386.21
## DLRSN ~ R10 + R7 + R9 + R8
##
## Df Deviance AIC
## + R2 1 2342.8 2354.8
## + R3 1 2370.4 2382.4
## + R4 1 2372.4 2384.4
## + R6 1 2373.5 2385.5
## <none> 2376.2 2386.2
## + R1 1 2375.1 2387.1
## + R5 1 2376.2 2388.2
## - R7 1 2401.7 2409.7
## - R8 1 2437.8 2445.8
## - R9 1 2438.7 2446.7
## - R10 1 2905.1 2913.1
##
## Step: AIC=2354.84
## DLRSN ~ R10 + R7 + R9 + R8 + R2
##
## Df Deviance AIC
## + R3 1 2330.3 2344.3
## + R6 1 2332.0 2346.0
## + R4 1 2340.8 2354.8
## <none> 2342.8 2354.8
## + R5 1 2341.9 2355.9
## + R1 1 2342.7 2356.7
## - R9 1 2369.2 2379.2
## - R2 1 2376.2 2386.2
## - R7 1 2381.4 2391.4
## - R8 1 2437.3 2447.3
## - R10 1 2879.6 2889.6
##
## Step: AIC=2344.34
## DLRSN ~ R10 + R7 + R9 + R8 + R2 + R3
##
## Df Deviance AIC
## + R6 1 2319.4 2335.4
## + R4 1 2328.3 2344.3
## <none> 2330.3 2344.3
## + R5 1 2329.4 2345.4
## + R1 1 2330.2 2346.2
## - R3 1 2342.8 2354.8
## - R8 1 2349.5 2361.5
## - R9 1 2364.2 2376.2
## - R2 1 2370.4 2382.4
## - R7 1 2370.5 2382.5
## - R10 1 2872.8 2884.8
##
## Step: AIC=2335.44
## DLRSN ~ R10 + R7 + R9 + R8 + R2 + R3 + R6
##
## Df Deviance AIC
## + R1 1 2309.7 2327.7
## <none> 2319.4 2335.4
## + R4 1 2318.2 2336.2
## + R5 1 2319.2 2337.2
## - R6 1 2330.3 2344.3
## - R3 1 2332.0 2346.0
## - R8 1 2336.5 2350.5
## - R7 1 2338.0 2352.0
## - R9 1 2345.9 2359.9
## - R2 1 2368.1 2382.1
## - R10 1 2798.2 2812.2
##
## Step: AIC=2327.69
## DLRSN ~ R10 + R7 + R9 + R8 + R2 + R3 + R6 + R1
##
## Df Deviance AIC
## <none> 2309.7 2327.7
## + R4 1 2308.9 2328.9
## + R5 1 2309.6 2329.6
## - R1 1 2319.4 2335.4
## - R3 1 2322.3 2338.3
## - R9 1 2325.3 2341.3
## - R8 1 2327.7 2343.7
## - R6 1 2330.2 2346.2
## - R7 1 2337.9 2353.9
## - R2 1 2361.2 2377.2
## - R10 1 2763.3 2779.3Backward_bic_selection <- step(Full_model, k = log(nrow(Bankruptcy_train)))## Start: AIC=2400.11
## DLRSN ~ R1 + R2 + R3 + R4 + R5 + R6 + R7 + R8 + R9 + R10
##
## Df Deviance AIC
## - R5 1 2308.9 2392.1
## - R4 1 2309.6 2392.7
## - R9 1 2315.9 2399.1
## <none> 2308.7 2400.1
## - R1 1 2317.8 2400.9
## - R3 1 2321.3 2404.4
## - R6 1 2326.6 2409.7
## - R8 1 2327.0 2410.1
## - R7 1 2331.4 2414.5
## - R2 1 2358.3 2441.4
## - R10 1 2559.7 2642.8
##
## Step: AIC=2392.06
## DLRSN ~ R1 + R2 + R3 + R4 + R6 + R7 + R8 + R9 + R10
##
## Df Deviance AIC
## - R4 1 2309.7 2384.5
## <none> 2308.9 2392.1
## - R1 1 2318.2 2393.0
## - R9 1 2320.5 2395.4
## - R3 1 2321.6 2396.4
## - R8 1 2327.3 2402.1
## - R6 1 2328.1 2402.9
## - R7 1 2331.9 2406.7
## - R2 1 2358.3 2433.1
## - R10 1 2642.8 2717.6
##
## Step: AIC=2384.51
## DLRSN ~ R1 + R2 + R3 + R6 + R7 + R8 + R9 + R10
##
## Df Deviance AIC
## <none> 2309.7 2384.5
## - R1 1 2319.4 2385.9
## - R3 1 2322.3 2388.8
## - R9 1 2325.3 2391.8
## - R8 1 2327.7 2394.2
## - R6 1 2330.2 2396.7
## - R7 1 2337.9 2404.4
## - R2 1 2361.2 2427.7
## - R10 1 2763.3 2829.8AIC - Backward Selection = 2327.69
AIC - Forward Selection = 2327.69
AIC - Both Selection = 2327.69
AIC - Backward BIC Selection = 2327.69
BIC - Backward Selection = 2384.51
BIC - Forward Selection = 2384.51
BIC - Both Selection = 2384.51
BIC - Backward BIC Selection = 2384.51
In-sample Prediction
# ROC Curve #
# In-sample prediction
pred_Bankruptcy_train <- predict(Forward_selection, type = "response")
library(ROCR)
pred <- prediction(pred_Bankruptcy_train, Bankruptcy_train$DLRSN)
perf <- performance(pred, "tpr", "fpr")
plot(perf, colorize = TRUE)
AUC for in-sample prediction = 0.87
Dimension for train data: 4077, 11
Dimension for test data: 1359, 11
Optimal Cut-off
# Optimal Cut-off
costfunc = function(obs, pred.p, pcut){
weight1 = 35 # define the weight for "true=1 but pred=0" (FN)
weight0 = 1 # define the weight for "true=0 but pred=1" (FP)
c1 = (obs == 1) & (pred.p < pcut) # count for "true=1 but pred=0" (FN)
c0 = (obs == 0) & (pred.p >= pcut) # count for "true=0 but pred=1" (FP)
cost = mean(weight1*c1 + weight0*c0) # misclassification with weight
return(cost) # you have to return to a value when you write R functions
}
p.seq = seq(0.01, 1, 0.01)
cost = rep(0, length(p.seq))
for (i in 1:length(p.seq)) {
cost[i] = costfunc(obs = Bankruptcy_train$DLRSN,
pred.p = pred_Bankruptcy_train, pcut = p.seq[i])
} # end of the loop
plot(p.seq, cost)
optimal.pcut = p.seq[which(cost == min(cost))]Optimal cut-off = 0.02
Contingency table for train data set
class_Bankruptcy_train_opt <- (pred_Bankruptcy_train > optimal.pcut)*1
table(Bankruptcy_train$DLRSN, class_Bankruptcy_train_opt,
dnn = c("True", "Predicted"))## Predicted
## True 0 1
## 0 1105 2384
## 1 11 577In_AMR <- (35*11+1*577)/4077Asymmetric Misclassification rate for train data = 0.24
Out-of-sample Prediction
pred_Bankruptcy_test <- predict(Forward_selection,
newdata = Bankruptcy_test, type = "response")
pred_test <- prediction(pred_Bankruptcy_test, Bankruptcy_test$DLRSN)
perf_test <- performance(pred_test, "tpr", "fpr")
plot(perf_test, colorize = TRUE)
AUC for out-of-sample prediction = 0.9
class_Bankruptcy_test_opt <- (pred_Bankruptcy_test > optimal.pcut)*1
table(Bankruptcy_test$DLRSN, class_Bankruptcy_test_opt,
dnn = c("True", "Predicted"))## Predicted
## True 0 1
## 0 360 811
## 1 2 186Out_AMR <- (35*2 + 1*186)/1359Asymmetric Misclassification rate on test data = 0.19
Executive Summary
Final Model
We developed the following models and calculated their AIC and BIC to find the best model. Stepwise – forward, backward and backward with BIC gave us the same result. We chose Stepwise – Backward with BIC as our final model.
- Model Type | Model | AIC | BIC
- Logistic Regression | DLRSN ~ R1 + R2 + R3 + R4 + R5 + R6 + R7 + R8 + R9 + R10 | 2330.66 | 2400.11
- Stepwise - Forward DLRSN ~ R10 + R7 + R9 + R8 + R2 + R3 + R6 + R1 2327.69 2384.51
- Stepwise - Backward DLRSN ~ R1 + R2 + R3 + R6 + R7 + R8 + R9 + R10 2327.69 2384.51
- Stepwise - Backward using BIC DLRSN ~ R1 + R2 + R3 + R6 + R7 + R8 + R9 + R10 2327.69 2384.51
- LASSO DLRSN ~ R1 + R2 + R3 + R4 + R5 + R6 + R7 + R8 + R9 + R10 -1033.94 -970.86
AUC and cost function were calculated for in sample and out of sample data. Following results were obtained:
Training Data
plot(perf, colorize = TRUE)
AUC for in-sample prediction = 0.87
Contingency table on train data
class_Bankruptcy_train_opt <- (pred_Bankruptcy_train > optimal.pcut)*1
table(Bankruptcy_train$DLRSN, class_Bankruptcy_train_opt,
dnn = c("True", "Predicted"))## Predicted
## True 0 1
## 0 1105 2384
## 1 11 577Asymmetric Misclassification rate for train data = 0.24
Test Data
plot(perf_test, colorize = TRUE)
AUC for out-of-sample prediction = 0.9
Contingency table on test data
table(Bankruptcy_test$DLRSN, class_Bankruptcy_test_opt,
dnn = c("True", "Predicted"))## Predicted
## True 0 1
## 0 360 811
## 1 2 186Asymmetric Misclassification rate on test data = 0.19
AUC for out of sample was found to be more. Cost function of out-of-sample was found to be less than in sample data.