Step 11 – Synoptic table of variable properties

We are using the bin.var() function for isopopulation “binning”

# Loading the bin.var function for encoding continuous variables bin.var <- function (x, bins = 4, method = c(“intervals”, “proportions”,
“natural”), labels = FALSE)
{
method <- match.arg(method)
if (length(x) < bins) {
stop(“The number of bins exceeds the number of data values”)
}
x <- if (method == “intervals”)
cut(x, bins, labels = labels)
else if (method == “proportions”)
cut(x, quantile(x, probs = seq(0, 1, 1/bins), na.rm = TRUE),
include.lowest = TRUE, labels = labels)
else {
xx <- na.omit(x)
breaks <- c(-Inf, tapply(xx, KMeans(xx, bins)$cluster,
max))
cut(x, breaks, labels = labels)
}
as.factor(x)
}

We also decide to process the distributions after removal of “extreme” outliers in order to avoid spurious relations And, we will not correct these values by interpolation.

wcs2train.ratios.NA <- wcs2train.ratios
for(i in 1:length(wcs2train.ratios.NA)){
# use the function to identify extreme outliers
extreme.outl <- FindOutliers(wcs2train.ratios.NA[,i])
# Replacing extreme outliers values by NA
wcs2train.ratios.NA[,i][extreme.outl] <- NA
cat(sprintf(“%s\n”, colnames(wcs2train.ratios)[i]))
}

Some variables are not amenable to “Binning” by the bin.var() function (with index going from 1:34):

• 8 – INVENTOR -> This variable has 154 values = -1
• 11 – V95A -> This variable is equal to INVENTOR/SALES, has 154 corresponding values = 0
• 18 – EXTRIC -> This variable should probably be excluded from this process. It has 502 values = 1 which are either convential values or excessive rounding operation results.
• 19 – TAXESONG -> This variable should probably excluded from this process. It is a similar case than EXTRIC. It has 305 values = 0 and 5 values = NA (A rare case in this datatable)
• 22 – V110A -> This variable is equal to INVENTOR/TOTAL ASSETS and has 154 corresponding values = 0
• 28 – SALESONV -> This variable should probably be excluded from this process. Its is equal to SALES/Value of production (or OPRE), and has 517 values = 100

We decide to retrain the variables: INVENTOR, V95A and V110A after masking their exception values by recoding as NA The variables: EXTRIC, TAXESONG and SALESONV are excluded from this study phase.

wcs2train.ratios.MON <- wcs2train.ratios.NA wcs2train.ratios.MON[[“INVENTOR”]] [ wcs2train.ratios.MON[[“INVENTOR”]] == -1 ] <- NA wcs2train.ratios.MON[[“V95A”]] [ wcs2train.ratios.MON[[“V95A”]] == 0 ] <- NA wcs2train.ratios.MON[[“V110A”]] [ wcs2train.ratios.MON[[“V110A”]] == 0 ] <- NA

The isopopulation “binning” avoids processing EXTRIC, TAXESONG and SALESONV by use of a seq object (idem, list of indices) It processe 31 variables.

STATUS <- wcs2train$BADGOOD RATIO9vSTATUS <- cbind.data.frame(STATUS=levels(STATUS)[STATUS]) varcount = 1 seq <- c(1:17,20:27,29:34) for(i in seq){
cat(sprintf(“%s – %s\n”, i, colnames(wcs2train.ratios.MON)[i]))
bins_cut = 9
for(j in 1:bins_cut) {
if(j == 1){
collist <- c(sprintf(“%s_1”,names(wcs2train.ratios.MON)[i]))
} else {
collist <- append(collist, sprintf(“%s_%s”,names(wcs2train.ratios.MON)[i],j))
}
}
ratio9 <- with(wcs2train.ratios.MON, bin.var(wcs2train.ratios.MON[,i], bins=bins_cut, method=’proportions’, labels=collist))
RATIO9vSTATUS <- cbind.data.frame(RATIO9vSTATUS, ratio9=levels(ratio9)[ratio9])
varcount = varcount + 1;
names(RATIO9vSTATUS)[varcount] = names(wcs2train.ratios.MON)[i]
}
ncol(RATIO9vSTATUS) [1] 32

This table has the variable STATUS in first column followed by the 31 “binned” variables. We finally generate out of this table presenting the default rates for 15 variables at time

# 15 figures arranged in 3 rows and 5 columns attach(mtcars) par(mfrow=c(3,5)) for(i in 2:16){
R9VB <- table(RATIO9vSTATUS[,i], RATIO9vSTATUS$STATUS)
badrate <- R9VB[,1]/(R9VB[,1]+R9VB[,2])*100.0
R9VB <- cbind(R9VB, badrate)
plot(R9VB[,3], main = sprintf(“Default rates for %s intervals”,names(RATIO9vSTATUS)[i]), xlab = sprintf(“%s(Binned)”,names(RATIO9vSTATUS)[i]), ylab=”Bad rate [%]”)
lines(R9VB[,3], type=”b”, pch=21, col=”black”, yaxt=”n”, lwd=2)
}

Default rates variation for variables: ROE,EBITDAON,ROI,ROA,V89A,ROS,ASSETSTU,INVENTOR,RECEIVAB,V94A,V95A,PAYABLES,COMMERCI,IEONEBIT,NIEONEBI (Table_4_26_Page 178_C2_Pt1.pdf)