R/fn_exp_categorical.R

In SmartEDA: Summarize and Explore the Data

#' Function to create frequency and custom tables
#'
#' this function will automatically select categorical variables and generate frequency or cross tables based on the user inputs. Output includes counts, percentages, row total and column total.
#'
#' @param data dataframe or matrix
#' @param Target target variable (dependent variable) if any. Default NULL
#' @param margin margin of index, 1 for row based proportions and 2 for column based proportions
#' @param clim maximum categories to be considered for frequency/custom table. Variables will be dropped if unique levels are higher than 'clim' for class factor/character variable. Default value is 10.
#' @param nlim numeric variable unique limits. Default 'nlim' values is 3, table excludes the numeric variables which is having greater than 'nlim' unique values
#' @param round round off
#' @param bin number of cuts for continuous target variable
#' @param per percentage values. Default table will give counts
#' @param weight a vector of weights, it must be equal to the length of data
#' @details this function provides both frequency and custom tables for all categorical features. And ouput will be generated in data frame
#'
#' @return Frequency tables, Cross tables
#'
#'  Columns description for frequency tables:
#'
#' \itemize{
#'   \item \code{Variable} is Variable name
#'   \item \code{Valid} is Variable values
#'   \item \code{Frequency} is Frequency
#'   \item \code{Percent} is Relative frequency
#'   \item \code{CumPercent} is Cumulative sum of relative frequency
#' }
#'
#'  Columns description for custom tables:
#'
#' \itemize{
#'   \item \code{Variable} is Variable name
#'   \item \code{Category} is Variable values
#'   \item \code{Count} is Number of counts
#'   \item \code{Per} is Percentages
#'   \item \code{Total} is Total count
#' }
#' @examples
#' # Frequency table
#' ExpCTable(mtcars, Target = NULL, margin = 1, clim = 10, nlim = 3, bin = NULL, per = FALSE)
#' # Crosstbale for Mtcars data
#' ExpCTable(mtcars, Target = "gear", margin = 1, clim = 10, nlim = 3, bin = NULL, per = FALSE)
#' # Weighted frequecncy for Mtcars data
#' ExpCTable(mtcars,  margin = 1, clim = 10, nlim = 3, bin = NULL, per = FALSE, weight = "wt")
#'
#' @importFrom stats na.omit complete.cases aggregate xtabs
#' @export ExpCTable

ExpCTable <- function(data, Target = NULL, margin = 1, clim = 10, nlim = 10, round = 2, bin = 3, per = FALSE, weight = NULL){
  r <- round
  if (!is.data.frame(data)) stop("Input data is not a dataframe")
  xx <- as.data.frame(data)
  sm_9xy_1wt8 <- NULL

  if (!is.null(weight)) {
    if(length(intersect(colnames(data), weight))>0) {
      if(!is.numeric(data[[weight]])) stop("Weight column is not numeric")
      sm_9xy_1wt8 = data[[weight]]
    } else
      if((length(weight) == nrow(data))) {
        sm_9xy_1wt8 = weight
      } else
        stop("length of weight vector should be equal to sample size 'n' OR specified weight column is not there in dataframe")
  }

  ### Variable selection
  num_var <- names(xx)[sapply(xx, is.numeric)]
  if (length(num_var) > 0){
    if (length(num_var) == 1) {
      xx1 <- as.data.frame(xx[, num_var])
      names(xx1) <- num_var
      } else {
        xx1 <- xx[, num_var]
      }
    num_var <- num_var[sapply(xx1, function(x){
      length(unique(na.omit(x))) > 1 & length(unique(na.omit(x))) <= nlim
      })]
  }
  Cat_v <- c(names(xx)[sapply(xx, is.character)], names(xx)[sapply(xx, is.factor)])
  if (length(Cat_v) + length(num_var) == 0) {
    message("Input data has no categorical input variables")
    return(Cat_v)
    }
  if (length(Cat_v) > 0){
    if (length(Cat_v) == 1) {
      xx1 <- as.data.frame(xx[, Cat_v]);
      names(xx1) <- Cat_v
      } else {
        xx1 <- xx[, Cat_v]
    }
    Cat_var <- Cat_v[sapply(xx1, function(x){
      length(unique(x)) <= clim & length(unique(x)) >= 2
      })]
    Cat_var <- Cat_var[!(Cat_var %in% Target)]
    Cat_var <- c(Cat_var, num_var)
  } else {
    Cat_var <- c(num_var)
  }
  if (length(Cat_var) <= 0) {
    message("Input data has no categorical input variables");
    return(Cat_var)
    }
  if (is.null(Target)) {
    Summary_FC <- data.frame()
    for (cat in Cat_var) {
      Xvar <- xx[, cat]
      Xvar <- as.factor(paste0(Xvar))
      nr <- nlevels(Xvar)
      FTable <- as.data.frame(matrix( numeric( (length(levels(Xvar)) + 1)),
                                     nrow = length(levels(Xvar)) + 1, ncol = 5))
      names(FTable) <- c("Variable", "Valid", "Frequency", "Percent", "CumPercent")

      if(is.null(weight)) {
        tbx <- table(Xvar)
        tbxp <- round(prop.table(table(Xvar)) * 100, r)
        tbxcp <- cumsum(tbxp)

        FTable$Variable <- cat
        FTable$Valid <- c(levels(Xvar), "TOTAL")
      } else {
        wt_table  = wtable(Xvar, sm_9xy_1wt8)
        tbx <- round(wt_table[[2]], 0)
        tbxp <- round(wt_table[[3]] * 100, r)
        tbxcp <- cumsum(tbxp)
        FTable$Variable <- cat
        FTable$Valid <- c(levels(Xvar), "TOTAL")
      }
      try(FTable[1 : nr, 3] <- tbx, silent = T)
      try(FTable[1 : nr, 4] <- tbxp, silent = T)
      try(FTable[1 : nr, 5] <- tbxcp, silent = T)
      FTable[nr + 1, "Frequency"] <- sum(tbx)
      FTable[nr + 1, "Percent"] <- NA
      FTable[nr + 1, "CumPercent"] <- NA
      Summary_FC <- rbind(Summary_FC, FTable)
    }
    class(Summary_FC) <- c("SmartEDA", "ExpCTable", "data.frame")
    return(Summary_FC)
  } else {
    Yvar <- xx[, Target]
    if (is.numeric(Yvar) & length(unique(Yvar)) > 5) {
      if (bin > nrow(xx)) stop("Number of bins are more than number of sample size")
      Yvar <- cut(Yvar, bin)
    } else Yvar <- as.factor(paste0(Yvar))

    if (length(Cat_var) >= 1){
      Summary_Cat <- data.frame ()
      for (cat in Cat_var) {
        Xvar <- xx[, cat]
        Xvar <- as.factor(paste0(Xvar))
        if (anyNA(Yvar)) Yvar <- addNA(Yvar)
        if (anyNA(Xvar)) Xvar <- addNA(Xvar)

        WTable <- as.data.frame(matrix( numeric( (length(levels(Xvar)) + 1) * ( (length(levels(Yvar))) + 3)),
                                       nrow = length(levels(Xvar)) + 1, ncol = (length(levels(Yvar))) + 3))
        names(WTable) <- c("VARIABLE", "CATEGORY", paste0(Target, ":", levels(Yvar)), "TOTAL")
        WTable$VARIABLE <- cat
        WTable$CATEGORY <- c(levels(Xvar), "TOTAL")
        nr <- length(levels(Xvar));
        nc <- length(levels(Yvar))
        tb <- table(Xvar, Yvar)
        tby <- prop.table(table(Yvar));
        tbx <- prop.table(table(Xvar))
        tby1 <- prop.table(table(Yvar), 1) * 100;
        tbx1 <- prop.table(table(Xvar), 1) * 100

        if (margin == 1){
                 tbl <- round(prop.table(tb, 2) * 100, r)
                 yvarper <- tby1; xvarper <- round(tbx * 100, r)
                 } else
        if (margin == 2){
                  tbl <- round(prop.table(tb, 1) * 100, r)
                 yvarper <- round(tby * 100, r); xvarper <- tbx1
                 }
        WTable_p <- WTable
        try(WTable[1 : nr, 3 : (nc + 2)] <- tb, silent = T)
        try(WTable_p[1 : nr, 3 : (nc + 2)] <- tbl, silent = T)
        csum <- apply(WTable[1 : nr, 3 : (nc + 2)], 2, sum)
        WTable[nr + 1, 3 : (nc + 2)] <- csum
        WTable_p[nr + 1, 3 : (nc + 2)] <- yvarper
        rsum <- apply(WTable[1 : (nr + 1), 3 : (nc + 2)], 1, sum)
        WTable[, "TOTAL"] <- rsum
        WTable_p[nr + 1, 3 : (nc + 2)] <- yvarper
        WTable_p["TOTAL"] <- c(xvarper, 100)
        if (per == TRUE) {
          WTable[, "Number"] <- "nn"
          WTable_p[, "Number"] <- "%"
          WTC <- rbind(WTable, WTable_p)
          nm <- names(WTC)
          ornam <- c("VARIABLE", "CATEGORY", "Number")
          ornam <- c(ornam, nm[ !nm %in% ornam])
          WTC <- WTC[, ornam]
        } else {
          WTC <- WTable
        }
        Summary_Cat <- rbind(Summary_Cat, WTC)
      }
      class(Summary_Cat) <- c("SmartEDA", "ExpCTable", "data.frame")
      return(Summary_Cat)
      } else stop("There is no categorical IV`s in data frame")
  }
}

#' Function provides summary statistics with weight of evidence
#'
#' Weight of evidence for categorical(X-independent) variable against Target variable (Y)
#'
#' @param X Independent categorical variable.
#' @param Y Binary response variable, it can take values of either 1 or 0.
#' @param valueOfGood Value of Y that is used as reference category.
#' @param print print results
#' @param Round rounds the values
#' @examples
#' X = mtcars$gear
#' Y = mtcars$am
#' WOE = ExpWoeTable(X,Y,valueOfGood = 1)
#' @seealso \code{\link[InformationValue:WOETable]{WOETable}}
#' @details
#' The weight of evidence tells the predictive power of an independent variable in relation to the dependent variable
#'
#' @return Weight of evidance summary table
#' @export ExpWoeTable

ExpWoeTable <- function (X, Y, valueOfGood = NULL, print = FALSE, Round = 2) {
  if (!is.factor(Y)) Y <- as.factor(paste0(Y))
  if (!is.factor(X)) X <- as.factor(paste0(X))
  if (anyNA(Y) == T) stop("Treat NA value from Target varible")
  if (anyNA(X)) X <- addNA(X)
  if (is.null(valueOfGood)) stop("Specify reference category for target variable")
  yClasses1 <- unique(Y)
  if (length(yClasses1) < 2) stop("Target variable having only one category")
  if (length(yClasses1) >= 2) {
    Y <- as.character(paste0(Y))
    yClasses <- unique(Y)
    if (valueOfGood == 1) {
      Y[which(Y == valueOfGood)] <- 1
      Y[which(!(Y == "1"))] <- 0
      } else {
        Y[which(Y == valueOfGood)] <- 99
        Y[which(!(Y == "99"))] <- 0
        Y[which(Y == 99)] <- 1
      }
    Y <- as.numeric(Y)
    woeTable <- as.data.frame(matrix(numeric(nlevels(X) * 10), nrow = nlevels(X), ncol = 10))
    names(woeTable) <- c("Class", "Out_1", "Out_0", "TOTAL",
                         "Per_1", "Per_0", "Odds_WOE", "Odds", "WOE", "IV")
    woeTable$Class <- levels(X)
    try(woeTable[, c(3, 2)] <- table(X, Y), silent = T)
    woeTable[, "TOTAL"] <- apply(woeTable[, c(3, 2)], 1, sum)
    tbx <- woeTable[, c(2, 3)]
    odsvalue <- NULL
    for (j in 1 : nrow(tbx)) {
      t <- sapply(rbind(tbx[j, ], apply(tbx[-j, ], 2, sum)), function(x){
        x[1] / x[2]
        })
      odv <- round(t[[1]] / t[[2]], Round)
      if (is.infinite(odv)) odv <- 0
      odsvalue <- rbind(odsvalue, c(t, od = odv))
    }
    woeTable$Per_1 <- round(woeTable$Out_1 / sum(woeTable$Out_1, na.rm = T), Round)
    woeTable$Per_0 <- round(woeTable$Out_0 / sum(woeTable$Out_0, na.rm = T), Round)
    woeTable$Odds_WOE <- round(woeTable$Per_1 / woeTable$Per_0, Round)
    woeTable[, "Odds"] <- odsvalue[, 3]
    woeTable$WOE <- round(log(woeTable$Odds_WOE), Round)
    woeTable$IV <- round( (woeTable$Per_1 - woeTable$Per_0) * woeTable$WOE, Round)
    woeTable[sapply(woeTable, is.infinite)] <- 0
    woeTable <- woeTable[, -7]
    attr(woeTable, "iValue") <- sum(woeTable$IV, na.rm = T)
    ref_1 <- valueOfGood
    ref_0 <- yClasses[!yClasses %in% valueOfGood]
    woeTable$Ref_1 <- paste0(ref_1)
    woeTable$Ref_0 <- paste0(ref_0, collapse = ", ")
    return(woeTable)
  }
}


#' Information value
#'
#' Provides information value for each categorical variable (X) against target variable (Y)
#'
#' @param X Independent categorical variable.
#' @param Y Binary response variable, it can take values of either 1 or 0.
#' @param valueOfGood Value of Y that is used as reference category.
#' @details
#' Information value is one of the most useful technique to select important variables in a predictive model. It helps to rank variables on the basis of their importance. The IV is calculated using the following formula
#' \itemize{
#'   \item \code{IV} - (Percentage of Good event - Percentage of Bad event) * WOE, where WOE is weight of evidence
#'   \item \code{WOE} - log(Percentage of Good event - Percentage of Bad event)
#' }
#'
#'Here is what the values of IV mean according to Siddiqi (2006)
#' \itemize{
#'   \item \code{If information value is < 0.03} then predictive power = "Not Predictive"
#'   \item \code{If information value is 0.03 to 0.1} then predictive power = "Somewhat Predictive"
#'   \item \code{If information value is 0.1 to 0.3} then predictive power = "Meidum Predictive"
#'   \item \code{If information value is >0.3} then predictive power = "Highly Predictive"
#' }
#'
#' @return Information value (iv) and Predictive power class
#' \itemize{
#'   \item \code{information} value
#'   \item \code{predictive} class
#' }
#'
#' @examples
#' X = mtcars$gear
#' Y = mtcars$am
#' ExpInfoValue(X,Y,valueOfGood = 1)
#' @export ExpInfoValue

ExpInfoValue <- function (X, Y, valueOfGood = NULL) {
  if (is.null(valueOfGood)) stop("Specify reference category for target variable")
  val <- valueOfGood
  woeTable <- ExpWoeTable(X = X, Y = Y, valueOfGood = val, print = FALSE)
  iv <- attr(woeTable, "iValue")
  if (is.nan(iv) | is.na(iv)) {
    PP <- "Not Predictive"
  }
  else if (iv < 0.09) {
    PP <- "Not Predictive"
  }
  else if (iv < 0.19) {
    PP <- "Somewhat Predictive"
  }
  else if (iv < 0.3) {
    PP <- "Medium Predictive"
  }
  else {
    PP <- "Highly Predictive"
  }
  return(list(`Information values` = iv, `Predictive class` = PP))
}


#' Function provides summary statistics for individual categorical predictors
#'
#' Provides bivariate summary statistics for all the categorical predictors against target variables. Output includes chi - square value, degrees of freedom, information value, p-value
#'
#' @param X Independent categorical variable.
#' @param Y Binary response variable, it can take values of either 1 or 0.
#' @param valueOfGood Value of Y that is used as reference category.
#' @details
#' Summary statistics included Pearson's Chi-squared Test for Count Data, "chisq.test" which performs chi-squared contingency table tests and goodness-of-fit tests. If any NA value present in X or Y variable, which will be considered as NA as in category while computing the contingency table.
#'
#' Also added unique levels for each X categorical variables and degrees of freedom
#'
#' @return The function provides summary statistics like
#'
#' \itemize{
#'   \item \code{Unique} number of levels
#'   \item \code{Chi square} statistics
#'   \item \code{P} value
#'   \item \code{df} Degrees of freedom
#'   \item \code{IV} Information value
#'   \item \code{Predictive} class
#' }
#'
#' @examples
#' X = mtcars$carb
#' Y = mtcars$am
#' ExpStat(X,Y,valueOfGood = 1)
#' @seealso \code{\link[stats:chisq.test]{chisq.test}}
#' @importFrom stats chisq.test
#' @export ExpStat

ExpStat <- function(X, Y, valueOfGood = NULL) {
  if (is.null(valueOfGood)) stop("Specify reference category for target variable")
  if (anyNA(Y)) stop("Treat NA value from Target varible")
  if (!inherits(Y, "factor")) Y <- as.factor(Y)
  if (!inherits(X, "factor")) X <- as.factor(X)
  tb <- table(X, Y)
  CTest <- chisq.test(tb, simulate.p.value = TRUE)
  ## Cramers V
  k <- min(dim(CTest$observed))
  N <- sum(CTest$observed)
  chi2 <- CTest$statistic
  CrV <- round(sqrt(chi2 / (N * (k - 1))), 2) #indicative of the degree of association
  if (is.nan(CrV) | is.na(CrV)) {
    Deg_asso <- "Very Weak"
    }
  else if (CrV < 0.09) Deg_asso <- "Very Weak"
  else if (CrV < 0.19) Deg_asso <- "Weak"
  else if (CrV < 0.3) Deg_asso <- "Moderate"
  else Deg_asso <- "Strong"
  TList <- rbind(
    Unique_Levels <- length(unique(X)),
    C2Tvalue <- round(CTest$statistic, 3),
    pval <- round(CTest$p.value, 3),
    df <- CTest$parameter,
    IVvalue <- round(ExpInfoValue(X, Y, valueOfGood = valueOfGood)[[1]], 3),
    Cram_Value <- CrV,
    Degree_Asso <- Deg_asso,
    Pred_class <- ExpInfoValue(X, Y, valueOfGood = valueOfGood)[[2]])
  return(TList)
}


#' Function provides summary statistics for all character or categorical columns in the dataframe
#'
#' This function combines results from weight of evidence, information value and summary statistics.
#'
#' @param data dataframe or matrix
#' @param Target target variable
#' @param result "Stat" - summary statistics, "IV" - information value
#' @param clim maximum unique levles for categorical variable. Variables will be dropped if unique levels is higher than clim for class factor/character variable
#' @param nlim maximum unique values for numeric variable.
#' @param bins number of bins (default is 10)
#' @param Pclass reference category of target variable
#' @param plot Information value barplot (default FALSE)
#' @param top for plotting top information values (default value is 20)
#' @param Round round of value
#' @details
#'Criteria used for categorical variable predictive power classification are

#' \itemize{
#'   \item \code{If information value is < 0.03} then predictive power = "Not Predictive"
#'   \item \code{If information value is 0.3 to 0.1} then predictive power = "Somewhat Predictive"
#'   \item \code{If information value is 0.1 to 0.3} then predictive power = "Meidum Predictive"
#'   \item \code{If information value is >0.3} then predictive power = "Highly Predictive"
#' }
#'
#' @return This function provides summary statistics for categorical variable
#' \itemize{
#'   \item \code{Stat} - Summary statistics includes Chi square test scores, p value, Information values, Cramers V and Degree if association
#'   \item \code{IV} - Weight of evidence and Information values
#' }
#'
#' Columns description:
#' \itemize{
#'   \item \code{Variable} variable name
#'   \item \code{Target} - Target variable
#'   \item \code{class} - name of bin (variable value otherwise)
#'   \item \code{out0} - number of good observations
#'   \item \code{out1} - number of bad observations
#'   \item \code{Total} - Total values for each category
#'   \item \code{pct1} - good observations / total good observations
#'   \item \code{pct0} - bad observations / total bad observations
#'   \item \code{odds} - Odds ratio [(a/b)/(c/d)]
#'   \item \code{woe} - Weight of Evidence – calculated as ln(odds)
#'   \item \code{iv} - Information Value  - ln(odds) * (pct0 – pct1)
#' }
#'
#' @examples
#' # Example 1
#' ## Read mtcars data
#' # Target variable "am" - Transmission (0 = automatic, 1 = manual)
#' # Summary statistics
#' ExpCatStat(mtcars,Target="am",result = "Stat",clim=10,nlim=10,bins=10,
#' Pclass=1,plot=FALSE,top=20,Round=2)
#' # Information value plot
#' ExpCatStat(mtcars,Target="am",result = "Stat",clim=10,nlim=10,bins=10,
#' Pclass=1,plot=TRUE,top=20,Round=2)
#' # Information value for categorical Independent variables
#' ExpCatStat(mtcars,Target="am",result = "IV",clim=10,nlim=10,bins=10,
#' Pclass=1,plot=FALSE,top=20,Round=2)
#' @author dubrangala
#' @importFrom scales wrap_format dollar_format
#' @export ExpCatStat


ExpCatStat <- function(data, Target=NULL, result="Stat", clim=10,
                       nlim=10, bins=10, Pclass=NULL, plot=FALSE, top=20, Round=2) {
  if (!is.data.frame(data)) stop("data must be a numeric vector or data.frame")
  if (is.null(Target)) stop("Target variable is missing")
  if (is.null(result)) stop("Specify the result option either 'Stat' or 'IV' for Information Value")
  result <- toupper(result)
  xx <- as.data.frame(data)
  Yvar <- as.factor(xx[, Target])
  if (length(levels(Yvar)) < 2) stop("Target variable has required atleast 2 categories")
  num_var <- names(xx)[sapply(xx, is.numeric)]
  Cat_var <- c(names(xx)[sapply(xx, is.character)], names(xx)[sapply(xx, is.factor)])

  if (length(num_var) > 0){
    num_var1 <- num_var[sapply(xx[, num_var], function(x){
      length(unique(na.omit(x))) > 1 & length(unique(na.omit(x))) <= nlim
      })]
    num_var2 <- num_var[sapply(xx[, num_var], function(x){
      length(unique(na.omit(x))) > nlim
      })]
  } else {
    num_var1 <- NULL
    num_var2 <- NULL
  }
  if ( (length(num_var1) + length(Cat_var) + length(num_var2)) == 0)
    stop("there is no categorical variable in the data")
  if (length(Cat_var) > 0) {
    Cat_varlst <- Cat_var[sapply(xx[, Cat_var], function(x) {
      length(unique(x)) <= clim & length(unique(x)) >= 2
      })]
    Cat_varlst <- c(Cat_varlst, num_var1)
    Cat_varlst <- Cat_varlst[!(Cat_varlst %in% Target)]
    } else {
    Cat_varlst <- num_var1[!(num_var1 %in% Target)]
  }
  if (is.null(Pclass)) {
    Pval <- 1
  } else {
      Pval <- Pclass
      }

  ## Adding Bin for Numeric variables
  if (length(num_var2) > 0 & !is.null(bins)) {
    if (bins > nrow(xx)) stop("number of bin is more than the number of rows in the dataframe")
    bin_data <- lapply(xx[, num_var2], function(x){
      q <- quantile(x, probs = c(1 : (bins - 1) / bins), na.rm = TRUE, type = 3)
      cuts <- unique(c(min(x), unique(q), max(x)))
      tp <- cut(x, cuts, include.lowest = T, dig.lab=6)
      return(tp)
    })
    bin_data <- cbind.data.frame(bin_data)
    inputdata <- cbind(subset(xx, select = Cat_varlst), bin_data)
  } else {
    inputdata <- xx[, Cat_varlst]
  }
  if (result == "STAT") {
    `IV Value` <- NULL
    Variable <- NULL
    tab2 <- sapply(inputdata, function(x){
      ExpStat(x, Yvar, valueOfGood = Pval)
      })
    tab2 <- data.frame(t(tab2))
    tb_op2 <- cbind(Var_name = rownames(tab2), Tar_v = Target, tab2)
    rownames(tb_op2) <- NULL
    tb_op2[, 3:8] <- sapply(tb_op2[, 3:8], function(x){
      as.numeric(paste0(x))
      })
    tb_op2[, -c(3:8)] <- sapply(tb_op2[, -c(3:8)], function(x){
      as.character(paste0(x))
      })
    names(tb_op2) <- c("Variable", "Target", "Unique", "Chi-squared", "p-value", "df", "IV Value", "Cramers V", "Degree of Association", "Predictive Power")
    class(tb_op2) <- c("SmartEDA", "data.frame")
    if (plot == TRUE) {
      varcnt <- nrow(tb_op2)
      if (top > varcnt) top <- varcnt
      plotdt <- tb_op2[order(tb_op2$`IV Value`, decreasing = T), ][1:top, ]
      gp_iv <- ggplot(plotdt, aes(y = `IV Value`,
                                  x = reorder(Variable, `IV Value`),
                                  label = paste0(`IV Value`))) +
        geom_bar(stat = "identity", position = "dodge", fill = "tan3") +
        xlab("Variables") +
        ylab("Information Value") +
        geom_text(size = 4, position = position_dodge(width = 0), vjust = 0.5, hjust = 0) +
        scale_x_discrete(labels = wrap_format(8)) +
        scale_y_continuous(labels = dollar_format(suffix = "", prefix = "")) +
        coord_flip() +
        theme(axis.line = element_line(size = 1, colour = "black"),
              panel.grid.major = element_line(colour = "#d3d3d3", linetype = "dashed"),
              panel.grid.minor = element_blank(),
              panel.border = element_blank(),
              panel.background = element_blank())
      print(gp_iv)
      return(tb_op2)
    }
    else return(tb_op2)
  }
  if (result == "IV") {
    tab1 <- lapply(inputdata, function(x) {
      ExpWoeTable (x, Yvar, valueOfGood = Pval, print = FALSE, Round = Round)
      })
    outp_dat <- do.call("rbind", tab1)
    outp_dat <- data.frame(Variable = rownames(outp_dat), outp_dat)
    outp_dat$Target <- Target
    rownames(outp_dat) <- NULL
    class(outp_dat) <- c("SmartEDA", "data.frame")
    return(outp_dat)
  }
  else stop("Input data error or Target variable is not properly defined")
}