R/html_transformation.R

In choonghyunryu/dlookr: Tools for Data Diagnosis, Exploration, Transformation

summary_imputation <- function(object, ...) {
  success <- attr(object, "success")
  
  if (!success) {
    message("imputation object isn't success.")
    return()
  }
  
  type <- attr(object, "type")
  method <- attr(object, "method")
  var_type <- attr(object, "var_type")
  
  original <- object
  
  if (type == "missing values") {
    na_pos <- attr(object, "na_pos")
    seed <- attr(object, "seed")
    
    original[na_pos] <- NA
  } else if (type == "outliers") {
    outlier_pos <- attr(object, "outlier_pos")
    outliers <- attr(object, "outliers")
    
    original[outlier_pos] <- outliers
  }
  
  if (var_type == "numerical") {
    original <- as.numeric(original)
    object <- as.numeric(object)
  } else if (var_type == "categorical") {
    original <- factor(original)
    object <- factor(object)
  }
  
  dframe <- data.frame(original = original,
                       imputation = object) %>%
    tidyr::gather()
  
  if (var_type == "numerical") {
    smmry <- dframe %>%
      group_by(key) %>%
      describe("value") %>%
      select(which(!names(.) %in% c("described_variables", "key"))) %>% 
      t
    
    smmry <- smmry[, 2:1]
    colnames(smmry) <- c("Original", "Imputation")
  } else if (var_type == "categorical") {
    tab_freq <- xtabs(~ value + key, dframe, addNA = TRUE)
    tab_relat <- round(prop.table(tab_freq, 2) * 100, 2)
    
    smmry <- cbind(tab_freq, tab_relat)
    smmry <- smmry[, c(2, 1, 4, 3)]
    colnames(smmry) <- c("original", "imputation",
                         "original_percent", "imputation_percent")
  }
  
  invisible(smmry)
}

summary_transform <- function(object, ...) {
  method <- attr(object, "method")
  origin <- attr(object, "origin")
  
  suppressWarnings({dframe <- data.frame(original = origin,
                                         trans = object) %>%
    tidyr::gather()})
  
  smmry <- dframe %>%
    group_by(key) %>%
    describe("value") %>%
    select(-described_variables, -key) %>%
    t
  colnames(smmry) <- c("Original", "Transformation")
  invisible(smmry)
}

#' @importFrom shiny tabsetPanel tabPanel
#' @importFrom htmltools plotTag
#' @import dplyr
#' @import reactable
html_impute_missing <- function(.data, target = NULL, base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  numerics <- c("mean", "median", "mode", "knn", "rpart", "mice")
  categories <- c("mode", "rpart", "mice")
  
  n_na <- .data %>% 
    purrr::map_dbl(function(x) sum(is.na(x))) %>% 
    .[. > 0]
  
  if (length(n_na) > 0) {
    data_type <- get_class(.data)
    
    tab_missing <- n_na %>% 
      tibble::as_tibble() %>% 
      mutate(variable = names(n_na)) %>% 
      rename("missing" = value) %>% 
      bind_cols(n = NROW(.data)) %>% 
      mutate(rate_missing = missing / n) %>% 
      inner_join(data_type, by = "variable") %>% 
      select(variable, class, n, missing, rate_missing) 
    
    cap <- "Information about impute missing values"
    html_cat(cap)
    
    tab_missing %>% 
      reactable(
        defaultColDef = colDef(style = "font-size: 14px;color: hsl(0, 0%, 40%);"),
        columns = list(
          variable = colDef(
            name = "Variables"
          ),        
          class = colDef(
            name = "Data Type"
          ),
          n = colDef(
            name = "Observations"
          ),
          missing = colDef(
            name = "Missing"
          ),          
          rate_missing = colDef(
            name = "Missing(%)",
            format = colFormat(
              percent = TRUE,
              digits = 2
            )
          )
        ),
        details = function(index) {
          variable <- tab_missing$variable[index]
          type <- tab_missing$class[index] %>% 
            as.character()
          
          if (type %in% c("integer", "numeric")) {
            method <- numerics
          } else if (type %in% c("factor", "ordered")) {
            method <- categories
          }
          
          if (is.null(target)) {
            method <- setdiff(method, c("knn", "rpart", "mice"))
          } else {
            method <- method
          }
          
          imputes <- method %>% 
            lapply(function(x) {
              if (is.null(target)) {
                imputate_na(.data, all_of(variable), all_of(target), method = x, 
                            print_flag = FALSE)
              } else {
                imputate_na(.data, all_of(variable),  method = x, 
                            print_flag = FALSE)
              }
            })
          
          p_compare <- imputes %>% 
            lapply(function(x) {
              htmltools::plotTag({
                plot(x, base_family = base_family)
              }, sprintf("A plot of imputation"), 
              width = 600, height = 400, device = grDevices::png)
            })
          
          suppressMessages({
            mat <- imputes %>% 
              purrr::map_dfc(function(x) {
                summary_imputation(x) %>% 
                  as.data.frame()
              })
          })

          if (type %in% c("integer", "numeric")) {
            drops <- seq(ncol(mat))[seq(ncol(mat)) %% 2 == 1][-1]
            mat <- mat[, -c(drops)]
            names(mat) <- c("original", paste("inpute", method, sep = "_"))
            
            tab_compare <- mat %>% 
              t() %>% 
              as.data.frame() %>% 
              select(mean, sd, IQR, p00, p25, p50,  p75, p100, skewness, kurtosis) %>% 
              reactable(
                defaultColDef = colDef(
                  style = "font-size: 14px;color: hsl(0, 0%, 40%);",
                  format = colFormat(
                    digits = 2
                  )
                ),              
                columns = list(
                  mean = colDef(
                    name = "Mean"
                  ),
                  sd = colDef(
                    name = "Standard Devation"
                  ),
                  p00 = colDef(
                    name = "Min"
                  ),  
                  p25 = colDef(
                    name = "Q1"
                  ),
                  p50 = colDef(
                    name = "Median"
                  ),
                  p75 = colDef(
                    name = "Q3"
                  ),
                  p100 = colDef(
                    name = "Max"
                  ),
                  skewness = colDef(
                    name = "Skewness"
                  ),
                  kurtosis = colDef(
                    name = "Kurtosis"
                  )                
                )
              )
            
            tabs <- lapply(seq(length(method) + 1), function(x) {
              if (x == 1) {
                shiny::tabPanel("Distribution", tab_compare,
                                hr(style = "border-top: 1px solid black;"),
                                style = "padding-top:5px; padding-bottom:25px;")
              } else {
                shiny::tabPanel(title = paste0("Plot_", method[x-1]),
                                p_compare[[x-1]],
                                hr(style = "border-top: 1px solid black;"),
                                style = "padding-top:5px; padding-bottom:25px;")
              }
              
            })
            
            do.call(shiny::tabsetPanel, c(tabs, id = "viz"))
          } else if (type %in% c("factor", "ordered")) {
            drops <- seq(ncol(mat))[seq(ncol(mat)) %% 4 != 2][-1]
            mat <- mat[, -c(drops)]
            names(mat) <- c("original", paste("inpute", method, sep = "_"))
            
            tab_compare <- mat %>% 
              t() %>% 
              as.data.frame()
            
            names(tab_compare)[ncol(tab_compare)] <- "<Missing>"
            
            tab_compare_rate <- tab_compare / rowSums(tab_compare)
            
            tab_compare <- tab_compare %>% 
              reactable(
                defaultColDef = colDef(
                  style = "font-size: 14px;color: hsl(0, 0%, 40%);",
                  format = colFormat(
                    separators = TRUE
                  )
                )
              )
            
            tab_compare_rate <- tab_compare_rate %>% 
              reactable(
                defaultColDef = colDef(
                  style = "font-size: 14px;color: hsl(0, 0%, 40%);",
                  format = colFormat(
                    percent = TRUE,
                    digits = 2
                  )
                )
              )
            
            tabs <- lapply(seq(length(method) + 2), function(x) {
              if (x == 1) {
                shiny::tabPanel("Contingency Table", tab_compare,
                                hr(style = "border-top: 1px solid black;"),
                                style = "padding-top:5px; padding-bottom:25px;")
              } else if (x == 2) {
                shiny::tabPanel("Relate Contingency Table", tab_compare_rate,
                                hr(style = "border-top: 1px solid black;"),
                                style = "padding-top:5px; padding-bottom:25px;")
              } else {              
                shiny::tabPanel(title = paste0("Plot_", method[x-2]),
                                p_compare[[x-2]],
                                hr(style = "border-top: 1px solid black;"),
                                style = "padding-top:5px; padding-bottom:25px;")
              }
              
            })
            
            do.call(shiny::tabsetPanel, c(tabs, id = "viz"))
          }  
        }  
      )
  } else {
    html_cat("There are no variables in the dataset with missing values.")
  }
}



#' @importFrom shiny tabsetPanel tabPanel
#' @importFrom htmltools plotTag
#' @import dplyr
#' @import reactable
html_impute_outlier <- function(.data, base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  method <- c("mean", "median", "mode", "capping")
  
  n <- nrow(.data)
  outlist <- find_outliers(.data, index = FALSE)
  
  if (length(outlist) > 0) {
    tab_outlier <- .data %>% 
      select_at(outlist) %>% 
      diagnose_numeric() %>% 
      mutate(n = n) %>% 
      mutate(rate_outlier = outlier / n) %>% 
      select(variables, n, min, max, outlier, rate_outlier) 
    
    cap <- "Information about impute outliers"
    html_cat(cap)
    
    tab_outlier %>% 
      reactable(
        defaultColDef = colDef(style = "font-size: 14px;color: hsl(0, 0%, 40%);"),
        columns = list(
          variables = colDef(
            name = "Variables"
          ),          
          n = colDef(
            name = "Observations"
          ),
          min = colDef(
            name = "Min"
          ), 
          max = colDef(
            name = "Max"
          ), 
          outlier = colDef(
            name = "Outlier"
          ),           
          rate_outlier = colDef(
            name = "Outlier(%)",
            format = colFormat(
              percent = TRUE,
              digits = 2
            )
          )
        ),
        details = function(index) {
          variable <- tab_outlier$variables[index]
          
          imputes <- method %>% 
            lapply(function(x) {
              imputate_outlier(.data, all_of(variable),  method = x)
            })
          
          p_compare <- imputes %>% 
            lapply(function(x) {
              htmltools::plotTag({
                plot(x, base_family = base_family)
              }, sprintf("A plot of imputation"), 
              width = 600, height = 400, device = grDevices::png)
            })
          
          suppressMessages({
            mat <- imputes %>% 
              purrr::map_dfc(function(x) {
                summary_imputation(x) %>% 
                  as.data.frame()
              })
          }) 
          
          drops <- seq(ncol(mat))[seq(ncol(mat)) %% 2 == 1][-1]
          mat <- mat[, -c(drops)]
          names(mat) <- c("original", paste("inpute", method, sep = "_"))
            
          tab_compare <- mat %>% 
            t() %>% 
            as.data.frame() %>% 
            select(mean, sd, IQR, p00, p25, p50,  p75, p100, skewness, kurtosis) %>% 
            reactable(
              defaultColDef = colDef(
                style = "font-size: 14px;color: hsl(0, 0%, 40%);",
                format = colFormat(
                  digits = 2
                )
              ),              
              columns = list(
                mean = colDef(
                  name = "Mean"
                ),
                sd = colDef(
                  name = "Standard Devation"
                ),
                p00 = colDef(
                  name = "Min"
                ),  
                p25 = colDef(
                  name = "Q1"
                ),
                p50 = colDef(
                  name = "Median"
                ),
                p75 = colDef(
                  name = "Q3"
                ),
                p100 = colDef(
                  name = "Max"
                ),
                skewness = colDef(
                  name = "Skewness"
                ),
                kurtosis = colDef(
                  name = "Kurtosis"
                )                
              )
            )
            
            tabs <- lapply(seq(length(method) + 1), function(x) {
              if (x == 1) {
                shiny::tabPanel("Distribution", tab_compare,
                                hr(style = "border-top: 1px solid black;"),
                                style = "padding-top:5px; padding-bottom:25px;")
              } else {
                shiny::tabPanel(title = paste0("Plot_", method[x-1]),
                                p_compare[[x-1]],
                                hr(style = "border-top: 1px solid black;"),
                                style = "padding-top:5px; padding-bottom:25px;")
              }
              
            })
            
            do.call(shiny::tabsetPanel, c(tabs, id = "viz"))
        }          
      )
  } else {
    html_cat("There are no variables in the dataset with outliers.")
  }
}


#' @importFrom shiny tabsetPanel tabPanel
#' @importFrom htmltools plotTag
#' @import dplyr
#' @import reactable
html_resolve_skewness <- function(.data, base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  skewlist <- find_skewness(.data, index = FALSE)
  
  if (length(skewlist) > 0) {
    tab_skewness <- .data %>% 
      select_at(skewlist) %>% 
      describe(statistics = c("quantiles", "skewness"),
               quantiles = c(0, 0.25, 0.5, 0.75, 1)) %>% 
      select(described_variables, p00:p100, skewness)
    
    cap <- "Information about resolve skewness"
    html_cat(cap)
    
    tab_skewness %>% 
      reactable(
        defaultColDef = colDef(
          style = "font-size: 14px;color: hsl(0, 0%, 40%);",
          minWidth = 80,
          maxWidth = 100
        ),
        columns = list(
          described_variables = colDef(
            name = "Variables",
            minWidth = 100,
            maxWidth = 150
          ),
          p00 = colDef(
            name = "Min"
          ),
          p25 = colDef(
            name = "Q1"
          ),
          p50 = colDef(
            name = "Median"
          ),
          p75 = colDef(
            name = "Q3"
          ),
          p100 = colDef(
            name = "Max"
          ),
          skewness = colDef(
            name = "Skewness",
            format = colFormat(
              digits = 2
            )
          )           
        ),
        details = function(index) {
          variable <- tab_skewness$described_variables[index]
          skewness <- tab_skewness$skewness[index] 
          
          if (skewness <= 0) {
            method <- c("1/x", "x^2", "x^3", "Box-Cox")
          } else {
            method <- c("log", "log+1", "sqrt", "Box-Cox")
          }
          
          resolve <- method %>% 
            lapply(function(x) {
              transform(pull(.data, variable),  method = x)
            })
          
          p_compare <- resolve %>% 
            lapply(function(x) {
              htmltools::plotTag({
                plot(x, base_family = base_family)
              }, sprintf("A plot of transform"), 
              width = 600, height = 400, device = grDevices::png)
            })
          
          suppressMessages({
            mat <- resolve %>% 
              purrr::map_dfc(function(x) {
                summary_transform(x) %>% 
                  as.data.frame()
              })
          }) 
          
          drops <- seq(ncol(mat))[seq(ncol(mat)) %% 2 == 1][-1]
          mat <- mat[, -c(drops)]
          names(mat) <- c("original", paste("transform", method, sep = "_"))
          
          tab_compare <- mat %>% 
            t() %>% 
            as.data.frame() %>% 
            select(mean, sd, IQR, p00, p25, p50,  p75, p100, skewness, kurtosis) %>% 
            reactable(
              defaultColDef = colDef(
                style = "font-size: 14px;color: hsl(0, 0%, 40%);",
                format = colFormat(
                  digits = 2
                )
              ),              
              columns = list(
                mean = colDef(
                  name = "Mean"
                ),
                sd = colDef(
                  name = "Standard Devation"
                ),
                p00 = colDef(
                  name = "Min"
                ),  
                p25 = colDef(
                  name = "Q1"
                ),
                p50 = colDef(
                  name = "Median"
                ),
                p75 = colDef(
                  name = "Q3"
                ),
                p100 = colDef(
                  name = "Max"
                ),
                skewness = colDef(
                  name = "Skewness"
                ),
                kurtosis = colDef(
                  name = "Kurtosis"
                )                
              )
            )
          
          tabs <- lapply(seq(length(method) + 1), function(x) {
            if (x == 1) {
              shiny::tabPanel("Distribution", tab_compare,
                              hr(style = "border-top: 1px solid black;"),
                              style = "padding-top:5px; padding-bottom:25px;")
            } else {
              shiny::tabPanel(title = paste0("Plot_", method[x-1]),
                              p_compare[[x-1]],
                              hr(style = "border-top: 1px solid black;"),
                              style = "padding-top:5px; padding-bottom:25px;")
            }
            
          })
          
          do.call(shiny::tabsetPanel, c(tabs, id = "viz"))
        }
      )
  } else {
    html_cat("There are no variables including skewed.")
  }
}  



#' @importFrom shiny tabsetPanel tabPanel
#' @importFrom htmltools plotTag
#' @importFrom purrr map_int
#' @import dplyr
#' @import reactable
html_binning <- function(.data, base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  method <- c("quantile", "equal", "pretty", "kmeans", "bclust")
  
  numlist <- find_class(.data, "numerical", index = FALSE)
  
  if (length(numlist) > 0) {
    cap <- "Binning information for numeric variables"
    html_cat(cap)
    
    tab_numerical <- .data %>% 
      select_at(numlist) %>% 
      diagnose() %>% 
      full_join(data.frame(method = method, stringsAsFactors = FALSE),
                by = character()) %>% 
      select(variables, types, unique_count, unique_rate, method)
    
    options(show.error.messages = FALSE)
    
    bins <- seq(nrow(tab_numerical)) %>% 
      lapply(function(x) {
        binn <- try(binning(pull(.data, tab_numerical$variables[x]), 
                            type = tab_numerical$method[x]),
                    silent = TRUE)
        if (inherits(binn, "try-error")) {
          return(NULL) 
        } else {
          return(binn)
        }  
      })
    
    # remove error cases
    is_complete <- sapply(bins, function(x) is(x)[1] == "bins")
    bins <- bins[is_complete]
    tab_numerical <- tab_numerical %>% 
      filter(is_complete)
      
    options(show.error.messages = TRUE)
    
    tab_numerical$n_bins <- bins %>% 
      purrr::map_int(function(x) attr(x, "levels") %>% length)
    
    tab_numerical %>% 
      reactable(
        defaultColDef = colDef(style = "font-size: 14px;color: hsl(0, 0%, 40%);"),
        columns = list(
          variables = colDef(
            name = "Variables",
            minWidth = 100,
            maxWidth = 150
          ),
          types = colDef(
            name = "Data Types",
            minWidth = 80,
            maxWidth = 100
          ),
          unique_count = colDef(
            name = "Unique",
            format = colFormat(
              separators = TRUE
            )
          ),
          unique_rate = colDef(
            name = "Unique Rate",
            format = colFormat(
              digits = 3
            )
          ),
          method = colDef(
            name = "Binning Method",
            minWidth = 100,
            maxWidth = 150
          ),
          n_bins = colDef(
            name = "Bins",
            minWidth = 50,
            maxWidth = 80
          )            
        ),
        details = function(index) {
          p_hist <- htmltools::plotTag({
            plot(bins[[index]], base_family = base_family)
            }, sprintf("A plot of bins"), 
            width = 600, height = 400, device = grDevices::png)
          
          tab_bins <- summary(bins[[index]]) %>% 
            reactable(
              defaultColDef = colDef(style = "font-size: 14px;color: hsl(0, 0%, 40%);"),
              columns = list(
                levels = colDef(
                  name = "Bins"
                ),
                freq = colDef(
                  name = "Frequency",
                  format = colFormat(
                    separators = TRUE
                  )
                ),
                rate = colDef(
                  name = "Frequency(%)",
                  format = colFormat(
                    percent = TRUE,
                    digits = 2
                  )
                )
              )
            )
          
          shiny::tabsetPanel(
            shiny::tabPanel("Distribution", p_hist,
                            hr(style = "border-top: 1px solid black;"),
                            style = "padding-top:5px; padding-bottom:25px;"), 
            shiny::tabPanel("Frequency Table", tab_bins,
                            hr(style = "border-top: 1px solid black;"),
                            style = "padding-top:5px; padding-bottom:25px;")
          )
        }
      )
  } else {
    html_cat("There are no numerical variables.")
  }
}  


#' @importFrom shiny tabsetPanel tabPanel
#' @importFrom htmltools plotTag
#' @importFrom purrr map_int
#' @import dplyr
#' @import reactable
html_optimal_binning <- function(.data, target, base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  numlist <- find_class(.data, "numerical", index = FALSE)
  
  if (is.null(target)) {
    html_cat("The target variable is not defied.")    
  } else if (!target %in% names(.data)) {
    html_cat(paste0("The target variable ", target, " is not in the data."))
  } else if (length(numlist) == 0) {
    html_cat("There are no numerical variables.")
  } else {
    n_levles <- length(table(pull(.data, target)))
    
    if (n_levles != 2) {
      html_cat("The target variable is not a binary class.")
    } else {
      tab_numerical <- .data %>% 
        select_at(numlist) %>% 
        diagnose() %>% 
        select(-missing_count, -missing_percent)
      
      # Optimal Binning for Scoring Modeling
      bins <- lapply(numlist, function(x)
        binning_by(.data, y = target, x = all_of(x), p = 0.05))
      
      success <- ifelse(sapply(bins, is.character),
                        "No significant splits", "Success")
      
      tab_numerical <- tab_numerical %>% bind_cols(
        data.frame(success = success))
        
      tab_numerical$n_bins <- bins %>% 
        purrr::map_int(function(x) {
          attr(x, "levels") %>% length
        })  
      
      cap <- "Optimal binning information for numeric variables with target variable"
      html_cat(cap)
      
      tab_numerical %>% 
        reactable(
          defaultColDef = colDef(style = "font-size: 14px;color: hsl(0, 0%, 40%);"),
          columns = list(
            variables = colDef(
              name = "Variables",
              minWidth = 100,
              maxWidth = 150
            ),
            types = colDef(
              name = "Data Types",
              minWidth = 80,
              maxWidth = 100
            ),
            unique_count = colDef(
              name = "Unique",
              format = colFormat(
                separators = TRUE
              )
            ),
            unique_rate = colDef(
              name = "Unique Rate",
              format = colFormat(
                digits = 3
              )
            ),
            success = colDef(
              name = "Success"
            ),
            n_bins = colDef(
              name = "Bins",
              minWidth = 50,
              maxWidth = 80
            )            
          ),
          details = function(index) {
            if (tab_numerical$n_bins[index] > 0) {
              p_dist <- htmltools::plotTag({
                plot(bins[[index]], base_family = base_family)
              }, sprintf("A plot of bins"), 
              width = 600, height = 400, device = grDevices::png)
              
              sticky_style <- list(position = "sticky", left = 0, 
                                   background = "#fff", zIndex = 1,
                                   borderRight = "1px solid #eee",
                                   fontSize = "15px")
              
              tab_bins <- attr(bins[[index]], "performance") %>% 
                select(-CntCumPos, -CntCumNeg, -RateCumPos, -RateCumNeg) %>% 
                reactable(
                  sortable = FALSE,
                  defaultColDef = colDef(
                    style = "font-size: 14px;color: hsl(0, 0%, 40%);",
                    format = colFormat(
                      digits = 3
                    )
                  ),
                  columns = list(
                    Bin = colDef(
                      style = sticky_style,
                      headerStyle = sticky_style
                    )
                  )
                )
              
              shiny::tabsetPanel(
                shiny::tabPanel("Distribution", p_dist,
                                hr(style = "border-top: 1px solid black;"),
                                style = "padding-top:5px; padding-bottom:25px;"), 
                shiny::tabPanel("Binning Table", tab_bins,
                                hr(style = "border-top: 1px solid black;"),
                                style = "padding-top:5px; padding-bottom:25px;")
              )
            }
          }          
        )  
    }  
  }
}  




#' @import dplyr
#' @importFrom knitr kable
#' @importFrom kableExtra kable_styling add_header_above
html_paged_impute_missing <- function(.data, target = NULL, full_width = TRUE, 
                                      font_size = 13, base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  numerics <- c("mean", "median", "mode", "knn", "rpart", "mice")
  categories <- c("mode", "rpart", "mice")
  
  n_na <- .data %>% 
    sapply(function(x) sum(is.na(x))) %>% 
    .[. > 0]
  
  if (length(n_na) > 0) {
    data_type <- get_class(.data)
    
    tab_missing <- n_na %>% 
      tibble::as_tibble() %>% 
      mutate(variable = names(n_na)) %>% 
      rename("missing" = value) %>% 
      bind_cols(n = NROW(.data)) %>% 
      mutate(rate_missing = missing / n) %>% 
      inner_join(data_type, by = "variable") %>% 
      select(variable, class, n, missing, rate_missing) 
    
    nm_cols <- c("variables", "data types", "observations", "missing", "missing(%)")
    
    caption <- "Information of missing values"
    
    tab_missing %>% 
      knitr::kable(format = "html", digits = 2, caption = caption,
                   col.names = nm_cols) %>% 
      kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                position = "left") %>%
      gsub("font-size: initial !important;",
           "font-size: 12px !important;", .) %>%        
      cat() 
    
    break_page_asis()
    
    for (i in seq(NROW(tab_missing))) {
      variable <- tab_missing$variable[i]
      type <- tab_missing$class[i] %>% 
        as.character()
      
      el <- div(h3(variable))
      cat(as.character(el))
      
      if (type %in% c("integer", "numeric")) {
        method <- numerics
      } else if (type %in% c("factor", "ordered")) {
        method <- categories
      }
      
      if (is.null(target)) {
        method <- setdiff(method, c("knn", "rpart", "mice"))
      } else {
        method <- method
      }
      
      imputes <- method %>% 
        lapply(function(x) {
          if (is.null(target)) {
            imputate_na(.data, all_of(variable), all_of(target), method = x, 
                        print_flag = FALSE)
          } else {
            imputate_na(.data, all_of(variable),  method = x, 
                        print_flag = FALSE)
          }
        })
      
      p_compare <- imputes %>% 
        lapply(function(x) {
            plot(x, base_family = base_family)
        })
      
      suppressMessages({
        mat <- imputes %>% 
          purrr::map_dfc(function(x) {
            summary_imputation(x) %>% 
              as.data.frame()
          })
      })
      
      if (type %in% c("integer", "numeric")) {
        drops <- seq(ncol(mat))[seq(ncol(mat)) %% 2 == 1][-1]
        mat <- mat[, -c(drops)]
        names(mat) <- c("original", paste("inpute", method, sep = "_"))
        
        nm_cols <- c("mean", "standard devation", "IQR", "min", "Q1", "median", "Q3",
                     "max", "skewness", "kurtosis")
        
        tab_compare <- mat %>% 
          t() %>% 
          as.data.frame() %>% 
          select(mean, sd, IQR, p00, p25, p50,  p75, p100, skewness, kurtosis) 
        
        for (j in seq(length(method))) {
          el <- div(h4(paste0(variable, " - ", method[j])))
          cat(as.character(el))
          
          print(p_compare[[j]])
          
          break_line_asis(1)
          
          caption <- paste0("Distribution with ", method[j], " method")
          
          tab_compare[c(1, j + 1), ] %>% 
            knitr::kable(format = "html", digits = 2, caption = caption,
                         col.names = nm_cols) %>% 
            kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                      position = "left") %>%
            gsub("font-size: initial !important;",
                 "font-size: 12px !important;", .) %>%              
            cat() 
          
          break_page_asis()
        }
      } else if (type %in% c("factor", "ordered")) {
        drops <- seq(ncol(mat))[seq(ncol(mat)) %% 4 != 2][-1]
        mat <- mat[, -c(drops)]
        names(mat) <- c("original", paste("inpute", method, sep = "_"))
        
        tab_compare <- mat %>% 
          t() %>% 
          as.data.frame()
        
        names(tab_compare)[ncol(tab_compare)] <- "<Missing>"
        
        tab_compare_rate <- tab_compare / rowSums(tab_compare)
        
        for (j in seq(length(method))) {
          el <- div(h4(paste0(variable, " - ", method[j])))
          cat(as.character(el))
          
          print(p_compare[[j]])
          
          break_line_asis(1)
          
          caption <- paste0("Contingency table with ", method[j], " method")
          
          header_above <- c(1, NCOL(tab_compare))
          names(header_above) <- c("imputation method", target)
          
          tab_compare[c(1, j + 1), ] %>% 
            knitr::kable(format = "html", caption = caption,
                         format.args = list(big.mark = ",")) %>% 
            kableExtra::add_header_above(header_above) %>% 
            kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                      position = "left") %>%
            gsub("font-size: initial !important;",
                 "font-size: 12px !important;", .) %>%              
            cat() 
          
          break_line_asis(1)
          
          caption <- paste0("Relate Contingency Table with ", method[j], " method")
          
          tab_compare_rate[c(1, j + 1), ] %>% 
            knitr::kable(format = "html", digits = 2, caption = caption) %>% 
            kableExtra::add_header_above(header_above) %>% 
            kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                      position = "left") %>%
            gsub("font-size: initial !important;",
                 "font-size: 12px !important;", .) %>%              
            cat() 
          
          break_page_asis()
        }
      }  
    }  
  } else {
    html_cat("There are no variables in the dataset with missing values.")
    break_page_asis()
  }
}


#' @import dplyr
#' @import reactable
#' @importFrom knitr kable
#' @importFrom kableExtra kable_styling add_header_above
html_paged_impute_outlier <- function(.data, full_width = TRUE, font_size = 13, 
                                      base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  method <- c("mean", "median", "mode", "capping")
  
  n <- nrow(.data)
  outlist <- find_outliers(.data, index = FALSE)
  
  if (length(outlist) > 0) {
    tab_outlier <- .data %>% 
      select_at(outlist) %>% 
      diagnose_numeric() %>% 
      mutate(n = n) %>% 
      mutate(rate_outlier = outlier / n) %>% 
      select(variables, n, min, max, outlier, rate_outlier) 
    
    nm_cols <- c("variables", "observations", "min", "max", "outlier", "outlier(%)")
    
    caption <- "Information of outliers"
    
    tab_outlier %>% 
      knitr::kable(format = "html", digits = 2, caption = caption,
                   col.names = nm_cols) %>% 
      kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                position = "left") %>%
      gsub("font-size: initial !important;",
           "font-size: 12px !important;", .) %>%         
      cat() 
    
    break_page_asis()
    
    for (i in seq(NROW(tab_outlier))) {
      variable <- tab_outlier$variables[i]
      
      el <- div(h3(variable))
      cat(as.character(el))
      
      imputes <- method %>% 
        lapply(function(x) {
          imputate_outlier(.data, all_of(variable),  method = x)
        })
      
      p_compare <- imputes %>% 
        lapply(function(x) {
            plot(x, base_family = base_family)
        })
      
      suppressMessages({
        mat <- imputes %>% 
          purrr::map_dfc(function(x) {
            summary_imputation(x) %>% 
              as.data.frame()
          })
      }) 
  
      drops <- seq(ncol(mat))[seq(ncol(mat)) %% 2 == 1][-1]
      mat <- mat[, -c(drops)]
      names(mat) <- c("original", paste("inpute", method, sep = "_"))
      
      tab_compare <- mat %>% 
        t() %>% 
        as.data.frame() %>% 
        select(mean, sd, IQR, p00, p25, p50,  p75, p100, skewness, kurtosis) 
      
      nm_cols <- c("mean", "standard devation", "IQR", "min", "Q1", "median", "Q3",
                   "max", "skewness", "kurtosis")
      
      for (j in seq(length(method))) {
        el <- div(h4(paste0(variable, " - ", method[j])))
        cat(as.character(el))
        
        print(p_compare[[j]])
        
        break_line_asis(1)
        
        caption <- paste0("Distribution with ", method[j], " method")
        
        tab_compare[c(1, j + 1), ] %>% 
          knitr::kable(format = "html", digits = 2, caption = caption,
                       col.names = nm_cols) %>% 
          kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                    position = "left") %>%
          gsub("font-size: initial !important;",
               "font-size: 12px !important;", .) %>%            
          cat() 
        
        break_page_asis()
      }
    } 
  } else {
    html_cat("There are no variables in the dataset with outliers.")
    break_page_asis()
  }
}


#' @import dplyr
#' @importFrom knitr kable
#' @importFrom kableExtra kable_styling 
html_paged_resolve_skewness <- function(.data, full_width = TRUE, font_size = 13,
                                        base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  skewlist <- find_skewness(.data, index = FALSE)
  
  if (length(skewlist) > 0) {
    tab_skewness <- .data %>% 
      select_at(skewlist) %>% 
      dlookr::describe(statistics = c("quantiles", "skewness"),
                       quantiles = c(0, 0.25, 0.5, 0.75, 1)) %>% 
      select(described_variables, p00:p100, skewness)
    
    nm_cols <- c("variables", "min", "Q1", "median", "Q3", "max", "skewness")
    
    caption <- "Information of skewness"
    
    tab_skewness %>% 
      knitr::kable(format = "html", digits = 2, caption = caption,
                   col.names = nm_cols) %>% 
      kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                position = "left") %>%
      gsub("font-size: initial !important;",
           "font-size: 12px !important;", .) %>%        
      cat() 
    
    break_page_asis()
    
    for (i in seq(NROW(tab_skewness))) {
      variable <- tab_skewness$described_variables[i]
      
      el <- div(h3(variable))
      cat(as.character(el))
      
      skewness <- tab_skewness$skewness[i] 
      
      if (skewness <= 0) {
        method <- c("1/x", "x^2", "x^3", "Box-Cox")
      } else {
        method <- c("log", "log+1", "sqrt", "Box-Cox")
      }
      
      resolve <- method %>% 
        lapply(function(x) {
          transform(pull(.data, variable),  method = x)
        })
      
      # p_compare <- resolve %>% 
      #   lapply(function(x) {
      #       plot(x)
      #   })
      
      suppressMessages({
        mat <- resolve %>% 
          purrr::map_dfc(function(x) {
            summary_transform(x) %>% 
              as.data.frame()
          })
      }) 
      
      drops <- seq(ncol(mat))[seq(ncol(mat)) %% 2 == 1][-1]
      mat <- mat[, -c(drops)]
      names(mat) <- c("original", paste("transform", method, sep = "_"))
      
      tab_compare <- mat %>% 
        t() %>% 
        as.data.frame() %>% 
        select(IQR, p00, p25, p50,  p75, p100, skewness, kurtosis)
      
      nm_cols <- c("IQR", "min", "Q1", "median", "Q3",
                   "max", "skewness", "kurtosis")
      
      for (j in seq(length(method))) {
        el <- div(h4(paste0(variable, " - ", method[j])))
        cat(as.character(el))
        
        plot(resolve[[j]], base_family = base_family)
        
        break_line_asis(1)
        
        caption <- paste0("Distribution with ", method[j], " method")
        
        tab_compare[c(1, j + 1), ] %>% 
          knitr::kable(format = "html", digits = 2, caption = caption,
                       col.names = nm_cols) %>% 
          kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                    position = "left") %>%
          gsub("font-size: initial !important;",
               "font-size: 12px !important;", .) %>%            
          cat() 
        
        break_page_asis()
      }
    }  
  } else {
    html_cat("There are no variables including skewed.")
    break_page_asis()
  }
}  


#' @import dplyr
#' @importFrom knitr kable
#' @importFrom kableExtra kable_styling 
html_paged_binning <- function(.data, full_width = TRUE, font_size = 13, 
                               base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  method <- c("quantile", "equal", "pretty", "kmeans", "bclust")
  
  numlist <- find_class(.data, "numerical", index = FALSE)
  
  if (length(numlist) > 0) {
    tab_numerical <- .data %>% 
      select_at(numlist) %>% 
      diagnose() %>% 
      full_join(data.frame(method = method, stringsAsFactors = FALSE),
                by = character()) %>% 
      select(variables, types, unique_count, unique_rate, method)
    
    options(show.error.messages = FALSE)
    
    bins <- seq(nrow(tab_numerical)) %>% 
      lapply(function(x) {
        binn <- try(binning(pull(.data, tab_numerical$variables[x]), 
                            type = tab_numerical$method[x]),
                    silent = TRUE)
        
        if (inherits(binn, "try-error")) {          
          return(NULL) 
        } else {
          return(binn)
        }  
      })
    
    options(show.error.messages = TRUE)
    
    tab_numerical$n_bins <- bins %>% 
      sapply(function(x) attr(x, "levels") %>% length)
    
    nm_cols <- c("variables", "data types", "unique", "unique rate", 
                 "binning method", "bins")
    
    caption <- "Information of binnings"
    
    print_tab(tab_numerical, n_rows = 30, add_row = 3, caption = caption, 
              full_width = TRUE, font_size = 13, col.names = nm_cols, 
              digits = 3, big_mark = TRUE)
    
    variable_before <- ""
    
    for (i in seq(NROW(tab_numerical))) {
      variable <- tab_numerical$variables[i]
      
      if (variable_before != variable) {
        el <- div(h3(variable))
        cat(as.character(el))
      }

      variable_before <- variable
      
      el <- div(h4(paste0(variable, " - ", tab_numerical[i, 5])))
      cat(as.character(el))
      
      plot(bins[[i]], base_family = base_family)
      
      break_line_asis(1)
      
      nm_cols <- c("bins", "frequency", "frequency(%)")
      
      caption <- "Contingency table with bins"
      
      summary(bins[[i]]) %>% 
        knitr::kable(format = "html", digits = 2, caption = caption,
                     col.names = nm_cols) %>% 
        kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                  position = "left") %>%
        gsub("font-size: initial !important;",
             "font-size: 12px !important;", .) %>%          
        cat() 
      
      break_page_asis()      
    }  
  } else {
    html_cat("There are no numerical variables.")
    break_page_asis()
  }
}  



#' @importFrom purrr map_int
#' @import dplyr
#' @importFrom knitr kable
#' @importFrom kableExtra kable_styling 
html_paged_optimal_binning <- function(.data, target, full_width = TRUE, 
                                       font_size = 13, base_family = NULL) {
  if (is.null(base_family)) {
    base_family <- "Roboto Condensed"
  }
  
  numlist <- find_class(.data, "numerical", index = FALSE)
  
  if (is.null(target)) {
    html_cat("The target variable is not defied.")    
    break_page_asis()
  } else if (!target %in% names(.data)) {
    html_cat(paste0("The target variable ", target, " is not in the data."))
    break_page_asis()
  } else if (length(numlist) == 0) {
    html_cat("There are no numerical variables.")
    break_page_asis()
  } else {
    n_levles <- length(table(pull(.data, target)))
    
    if (n_levles != 2) {
      html_cat("The target variable is not a binary class.")
      break_page_asis()
    } else {
      tab_numerical <- .data %>% 
        select_at(numlist) %>% 
        diagnose() %>% 
        select(-missing_count, -missing_percent)
      
      # Optimal Binning for Scoring Modeling
      bins <- lapply(numlist, function(x)
        binning_by(.data, y = target, x = all_of(x), p = 0.05))
      
      success <- ifelse(sapply(bins, is.character),
                        "No significant splits", "Success")
      
      tab_numerical <- tab_numerical %>% bind_cols(
        data.frame(success = success))
      
      tab_numerical$n_bins <- bins %>% 
        purrr::map_int(function(x) {
          attr(x, "levels") %>% length
        })  
      
      nm_cols <- c("variables", "data types", "unique", "unique rate", 
                   "success", "bins")
      
      caption <- "Information of optimal binnings"
      
      print_tab(tab_numerical, n_rows = 30, add_row = 3, caption = caption, 
                full_width = TRUE, font_size = 13, col.names = nm_cols, 
                digits = 3, big_mark = TRUE)
      
      for (i in seq(NROW(tab_numerical))) {
        if (tab_numerical$n_bins[i] == 0) {
          next
        }
        
        variable <- tab_numerical$variables[i]

        el <- div(h3(variable))
        cat(as.character(el))
        
        plot(bins[[i]], base_family = base_family)
        
        break_line_asis(1)
        
        caption <- "Binning table with performance measures"
        
        attr(bins[[i]], "performance") %>% 
          select(-CntCumPos, -CntCumNeg, -RateCumPos, -RateCumNeg) %>%
          knitr::kable(format = "html", digits = 2, caption = caption) %>% 
          kableExtra::kable_styling(full_width = full_width, font_size = font_size, 
                                    position = "left") %>%
          gsub("font-size: initial !important;",
               "font-size: 12px !important;", .) %>%            
          cat() 
          
        break_page_asis()
      }  
    }  
  }
}