Clase práctica: Tablas de frecuencia, gráficos y medidas descriptivas (FIX)

Documento robusto para RPubs. Si df existe pero no es data frame, se ignora y se lee el Excel.
Nombres de columnas limpiados sin depender de janitor::make_clean_names(case=...).

0.1 0) Paquetes e importación

req <- c("dplyr","ggplot2","tidyr","scales","stringi","rlang","readxl")
to_install <- setdiff(req, rownames(installed.packages()))
if (length(to_install)) install.packages(to_install, dependencies = TRUE)

library(dplyr)
library(ggplot2)
library(tidyr)
library(scales)
library(stringi)
library(rlang)
library(readxl)

0.1.1 Importar datos y limpiar nombres

# Función de limpieza de nombres sin janitor (compatible 100%)
clean_names_vec <- function(x) {
  x <- as.character(x)
  # quitar acentos y caracteres latinos
  x <- stringi::stri_trans_general(x, "Latin-ASCII")
  # a minúsculas
  x <- tolower(x)
  # reemplazar no alfanum por _
  x <- gsub("[^a-z0-9]+", "_", x, perl = TRUE)
  # colapsar múltiples _
  x <- gsub("_+", "_", x, perl = TRUE)
  # quitar _ al inicio/fin
  x <- gsub("^_+|_+$", "", x, perl = TRUE)
  x
}

# Determinar origen de df
use_env_df <- exists("df")
if (use_env_df) {
  # si existe, confirmar que es data.frame/tibble
  if (!is.data.frame(df)) {
    message("⚠️ 'df' existe pero no es data.frame; se leerá desde Excel.")
    use_env_df <- FALSE
  }
}

if (!use_env_df) {
  excel_path <- "2016-2019 valores nominales.xlt (1).xlsx"
  if (!file.exists(excel_path)) {
    stop("No se encontró el archivo Excel. Ajuste 'excel_path' al path correcto.")
  }
  df <- readxl::read_excel(excel_path, sheet = "datos_originales")
}

# Forzar a data.frame clásico por si vienen clases raras
df <- as.data.frame(df, stringsAsFactors = FALSE)

# Limpiar nombres de columnas
colnames(df) <- clean_names_vec(colnames(df))

# Normalizar año -> anio
if ("ano" %in% names(df)) names(df)[names(df) == "ano"] <- "anio"
if (!"anio" %in% names(df) && "año" %in% names(df)) names(df)[names(df) == "año"] <- "anio"

# Unificar bancos_privatizados si viniera duplicada
if (!"bancos_privatizados" %in% names(df)) {
  cand <- grep("^bancos_privatizados", names(df), value = TRUE)
  if (length(cand) >= 1) names(df)[names(df) == cand[1]] <- "bancos_privatizados"
}

# Recode de factores útiles (si son 0/1 numéricos)
if ("reestructuracion" %in% names(df) && is.numeric(df$reestructuracion)) {
  df$reestructuracion <- factor(df$reestructuracion, levels = c(0,1), labels = c("No","Si"))
}
if ("bancos_privatizados" %in% names(df) && is.numeric(df$bancos_privatizados)) {
  df$bancos_privatizados <- factor(df$bancos_privatizados, levels = c(0,1), labels = c("No","Si"))
}

dplyr::glimpse(df)

## Rows: 96
## Columns: 8
## $ provincias          <chr> "Buenos Aires", "CABA", "Catamarca", "Chaco", "Chu…
## $ reestructuracion    <fct> Si, No, Si, Si, Si, Si, Si, Si, Si, Si, No, Si, Si…
## $ anio                <dbl> 2016, 2016, 2016, 2016, 2016, 2016, 2016, 2016, 20…
## $ resultado           <dbl> -17181.554432, -10143.663669, 9.304292, -4924.7512…
## $ deudanac            <dbl> 185976407.8, 11146195.3, 2495033.4, 12310491.0, 79…
## $ deudaext            <dbl> 11020102.17, 31093595.77, 124885.99, 737297.01, 12…
## $ deseq_vertical      <dbl> 0.3997638, 0.2306568, 0.8207827, 0.7350382, 0.3692…
## $ bancos_privatizados <fct> No, No, Si, No, No, No, No, Si, Si, Si, No, No, Si…

0.2 1) Tablas de distribución de frecuencias (cualitativas)

tabla_freq_cuali <- function(data, var) {
  var_quo <- rlang::enquo(var)
  data %>%
    mutate(.var = !!var_quo) %>%
    count(.var, name = "freq") %>%
    mutate(
      pct       = 100 * freq / sum(freq),
      freq_acum = cumsum(freq),
      pct_acum  = cumsum(pct)
    ) %>%
    rename(!!rlang::as_name(var_quo) := .var) %>%
    arrange(desc(freq))
}

if ("reestructuracion" %in% names(df)) {
  tab_reestr <- tabla_freq_cuali(df, reestructuracion)
  tab_reestr
}

if ("bancos_privatizados" %in% names(df)) {
  tab_ban <- tabla_freq_cuali(df, bancos_privatizados)
  tab_ban
}

if ("provincias" %in% names(df)) {
  tab_prov <- df %>%
    count(provincias, name = "freq") %>%
    mutate(pct = 100*freq/sum(freq)) %>%
    arrange(desc(freq))
  head(tab_prov, 10)
}

0.3 2) Gráficos de torta y barras (cualitativas)

grafico_torta <- function(tabla_cuali, var_lab = "Categoría") {
  primero <- names(tabla_cuali)[1]
  ggplot(tabla_cuali, aes(x = "", y = pct, fill = .data[[primero]])) +
    geom_col(width = 1) +
    coord_polar(theta = "y") +
    geom_text(aes(label = paste0(round(pct,1),"%")),
              position = position_stack(vjust = 0.5), size = 3) +
    labs(x = NULL, y = NULL, fill = var_lab, title = paste0("Gráfico de sectores — ", var_lab)) +
    theme_void()
}

grafico_barras <- function(tabla_cuali, var_lab = "Categoría") {
  primero <- names(tabla_cuali)[1]
  ggplot(tabla_cuali, aes(x = .data[[primero]], y = pct)) +
    geom_col() +
    geom_text(aes(label = paste0(round(pct,1),"%")), vjust = -0.4, size = 3) +
    labs(x = var_lab, y = "%", title = paste0("Barras por % — ", var_lab)) +
    theme_minimal()
}

if (exists("tab_reestr")) {
  grafico_torta(tab_reestr, "Reestructuración")
  grafico_barras(tab_reestr, "Reestructuración")
}

if (exists("tab_ban")) {
  grafico_torta(tab_ban, "Bancos privatizados")
  grafico_barras(tab_ban, "Bancos privatizados")
}

0.4 3) Distribuciones para cuantitativas: clases, histograma y medidas

tabla_freq_cuanti <- function(data, var, k = NULL) {
  x <- dplyr::pull(data, {{var}})
  x <- x[!is.na(x)]
  if (length(x) < 2) stop("La variable no tiene suficientes datos numéricos.")
  if (is.null(k)) k <- nclass.Sturges(x)
  brks <- seq(min(x), max(x), length.out = k + 1)
  clase <- cut(x, breaks = brks, include.lowest = TRUE, right = TRUE)
  tb <- as.data.frame(table(clase))
  names(tb) <- c("clase", "freq")
  tb <- tb %>%
    mutate(
      rel      = freq / sum(freq),
      pct      = 100*rel,
      acum     = cumsum(freq),
      rel_acum = cumsum(rel),
      pct_acum = 100*rel_acum
    )
  lims <- do.call(rbind, strsplit(gsub("\\[|\\]|\\(|\\)", "", as.character(tb$clase)), ",")) %>%
    as.data.frame()
  names(lims) <- c("li","ls")
  lims$li <- as.numeric(lims$li); lims$ls <- as.numeric(lims$ls)
  tb <- dplyr::bind_cols(tb, lims) %>%
    mutate(pm = (li + ls)/2)
  tb
}

resumen_num <- function(x) {
  x <- x[!is.na(x)]
  stopifnot(is.numeric(x))
  n  <- length(x)
  media   <- mean(x); mediana <- median(x)
  q <- as.numeric(quantile(x, probs = c(0, .25, .5, .75, 1)))
  names(q) <- c("min","Q1","Q2_mediana","Q3","max")
  rango <- diff(range(x))
  var_  <- var(x); sd_ <- sd(x); iqr_ <- IQR(x)
  cv    <- ifelse(abs(media) > .Machine$double.eps, sd_/abs(media), NA_real_)
  moda <- NA
  if (n > 0) {
    tab <- sort(table(x), decreasing = TRUE)
    moda <- as.numeric(names(tab)[1])
  }
  tibble::tibble(
    n = n, media = media, mediana = mediana,
    min = q[1], Q1 = q[2], Q2 = q[3], Q3 = q[4], max = q[5],
    rango = rango, varianza = var_, desvio = sd_, IQR = iqr_, CV = cv, moda = moda
  )
}

grafico_hist <- function(data, var, bins = NULL) {
  x <- dplyr::pull(data, {{var}})
  if (!is.numeric(x)) stop("La variable debe ser numérica para histograma.")
  if (is.null(bins)) bins <- nclass.Sturges(x)
  ggplot(data, aes(x = {{var}})) +
    geom_histogram(bins = bins) +
    labs(x = rlang::as_name(rlang::enquo(var)), y = "Frecuencia",
         title = paste0("Histograma — ", rlang::as_name(rlang::enquo(var)))) +
    theme_minimal()
}

vars_num <- c("resultado","deudanac","deudaext","deseq_vertical")
vars_num <- vars_num[vars_num %in% names(df)]

for (v in vars_num) {
  cat("\n\n### Variable:", v, "\n")
  print(tabla_freq_cuanti(df, !!sym(v)))
  print(resumen_num(dplyr::pull(df, !!sym(v))))
  print(grafico_hist(df, !!sym(v)))
}

## 
## 
## ### Variable: resultado 
##                   clase freq        rel       pct acum   rel_acum   pct_acum
## 1 [-1.72e+04,-9.83e+03]    5 0.05208333  5.208333    5 0.05208333   5.208333
## 2 (-9.83e+03,-2.48e+03]   24 0.25000000 25.000000   29 0.30208333  30.208333
## 3  (-2.48e+03,4.87e+03]   56 0.58333333 58.333333   85 0.88541667  88.541667
## 4   (4.87e+03,1.22e+04]    7 0.07291667  7.291667   92 0.95833333  95.833333
## 5   (1.22e+04,1.96e+04]    3 0.03125000  3.125000   95 0.98958333  98.958333
## 6   (1.96e+04,2.69e+04]    0 0.00000000  0.000000   95 0.98958333  98.958333
## 7   (2.69e+04,3.43e+04]    0 0.00000000  0.000000   95 0.98958333  98.958333
## 8   (3.43e+04,4.16e+04]    1 0.01041667  1.041667   96 1.00000000 100.000000
##       li    ls     pm
## 1 -17200 -9830 -13515
## 2  -9830 -2480  -6155
## 3  -2480  4870   1195
## 4   4870 12200   8535
## 5  12200 19600  15900
## 6  19600 26900  23250
## 7  26900 34300  30600
## 8  34300 41600  37950
## # A tibble: 1 × 14
##       n media mediana     min     Q1    Q2    Q3    max  rango  varianza desvio
##   <int> <dbl>   <dbl>   <dbl>  <dbl> <dbl> <dbl>  <dbl>  <dbl>     <dbl>  <dbl>
## 1    96 -52.7   -284. -17182. -2896. -284. 1587. 41613. 58794. 45059086.  6713.
## # ℹ 3 more variables: IQR <dbl>, CV <dbl>, moda <dbl>

## 
## 
## ### Variable: deudanac 
##                 clase freq        rel       pct acum  rel_acum  pct_acum
## 1  [2.7e+05,3.08e+07]   88 0.91666667 91.666667   88 0.9166667  91.66667
## 2 (3.08e+07,6.14e+07]    3 0.03125000  3.125000   91 0.9479167  94.79167
## 3  (6.14e+07,9.2e+07]    1 0.01041667  1.041667   92 0.9583333  95.83333
## 4  (9.2e+07,1.23e+08]    2 0.02083333  2.083333   94 0.9791667  97.91667
## 5 (1.23e+08,1.53e+08]    0 0.00000000  0.000000   94 0.9791667  97.91667
## 6 (1.53e+08,1.84e+08]    0 0.00000000  0.000000   94 0.9791667  97.91667
## 7 (1.84e+08,2.14e+08]    1 0.01041667  1.041667   95 0.9895833  98.95833
## 8 (2.14e+08,2.45e+08]    1 0.01041667  1.041667   96 1.0000000 100.00000
##         li       ls        pm
## 1 2.70e+05 3.08e+07  15535000
## 2 3.08e+07 6.14e+07  46100000
## 3 6.14e+07 9.20e+07  76700000
## 4 9.20e+07 1.23e+08 107500000
## 5 1.23e+08 1.53e+08 138000000
## 6 1.53e+08 1.84e+08 168500000
## 7 1.84e+08 2.14e+08 199000000
## 8 2.14e+08 2.45e+08 229500000
## # A tibble: 1 × 14
##       n  media mediana    min     Q1     Q2     Q3    max  rango varianza desvio
##   <int>  <dbl>   <dbl>  <dbl>  <dbl>  <dbl>  <dbl>  <dbl>  <dbl>    <dbl>  <dbl>
## 1    96 1.65e7  8.12e6 2.70e5 4.80e6 8.12e6 1.27e7 2.45e8 2.44e8  1.15e15 3.39e7
## # ℹ 3 more variables: IQR <dbl>, CV <dbl>, moda <dbl>

## 
## 
## ### Variable: deudaext 
##                 clase freq        rel       pct acum  rel_acum  pct_acum
## 1        [0,6.84e+07]   90 0.93750000 93.750000   90 0.9375000  93.75000
## 2 (6.84e+07,1.37e+08]    3 0.03125000  3.125000   93 0.9687500  96.87500
## 3 (1.37e+08,2.05e+08]    1 0.01041667  1.041667   94 0.9791667  97.91667
## 4 (2.05e+08,2.74e+08]    0 0.00000000  0.000000   94 0.9791667  97.91667
## 5 (2.74e+08,3.42e+08]    0 0.00000000  0.000000   94 0.9791667  97.91667
## 6  (3.42e+08,4.1e+08]    1 0.01041667  1.041667   95 0.9895833  98.95833
## 7  (4.1e+08,4.79e+08]    0 0.00000000  0.000000   95 0.9895833  98.95833
## 8 (4.79e+08,5.47e+08]    1 0.01041667  1.041667   96 1.0000000 100.00000
##         li       ls        pm
## 1 0.00e+00 6.84e+07  34200000
## 2 6.84e+07 1.37e+08 102700000
## 3 1.37e+08 2.05e+08 171000000
## 4 2.05e+08 2.74e+08 239500000
## 5 2.74e+08 3.42e+08 308000000
## 6 3.42e+08 4.10e+08 376000000
## 7 4.10e+08 4.79e+08 444500000
## 8 4.79e+08 5.47e+08 513000000
## # A tibble: 1 × 14
##       n   media mediana   min     Q1     Q2     Q3    max  rango varianza desvio
##   <int>   <dbl>   <dbl> <dbl>  <dbl>  <dbl>  <dbl>  <dbl>  <dbl>    <dbl>  <dbl>
## 1    96  2.33e7  1.86e6     0 4.26e5 1.86e6 1.99e7 5.47e8 5.47e8  4.95e15 7.04e7
## # ℹ 3 more variables: IQR <dbl>, CV <dbl>, moda <dbl>

## 
## 
## ### Variable: deseq_vertical 
##           clase freq        rel       pct acum   rel_acum   pct_acum    li
## 1 [0.208,0.295]    8 0.08333333  8.333333    8 0.08333333   8.333333 0.208
## 2 (0.295,0.381]    3 0.03125000  3.125000   11 0.11458333  11.458333 0.295
## 3 (0.381,0.468]    9 0.09375000  9.375000   20 0.20833333  20.833333 0.381
## 4 (0.468,0.555]   17 0.17708333 17.708333   37 0.38541667  38.541667 0.468
## 5 (0.555,0.641]   12 0.12500000 12.500000   49 0.51041667  51.041667 0.555
## 6 (0.641,0.728]   18 0.18750000 18.750000   67 0.69791667  69.791667 0.641
## 7 (0.728,0.815]   15 0.15625000 15.625000   82 0.85416667  85.416667 0.728
## 8 (0.815,0.901]   14 0.14583333 14.583333   96 1.00000000 100.000000 0.815
##      ls     pm
## 1 0.295 0.2515
## 2 0.381 0.3380
## 3 0.468 0.4245
## 4 0.555 0.5115
## 5 0.641 0.5980
## 6 0.728 0.6845
## 7 0.815 0.7715
## 8 0.901 0.8580
## # A tibble: 1 × 14
##       n media mediana   min    Q1    Q2    Q3   max rango varianza desvio   IQR
##   <int> <dbl>   <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>    <dbl>  <dbl> <dbl>
## 1    96 0.611   0.631 0.208 0.491 0.631 0.753 0.901 0.693   0.0347  0.186 0.262
## # ℹ 2 more variables: CV <dbl>, moda <dbl>

0.5 4) Resúmenes agrupados por año

if ("anio" %in% names(df)) {
  df %>%
    group_by(anio) %>%
    summarise(
      n = n(),
      media_resultado   = if ("resultado" %in% names(df)) mean(resultado, na.rm = TRUE) else NA_real_,
      mediana_resultado = if ("resultado" %in% names(df)) median(resultado, na.rm = TRUE) else NA_real_
    )
}

0.6 5) Pruebas simples e intervalos de confianza

# χ²: reestructuracion vs bancos_privatizados (si existen)
if (all(c("reestructuracion","bancos_privatizados") %in% names(df))) {
  print(chisq.test(table(df$reestructuracion, df$bancos_privatizados)))
}

## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  table(df$reestructuracion, df$bancos_privatizados)
## X-squared = 0.51429, df = 1, p-value = 0.4733

# t-test de medias: resultado por bancos_privatizados
if (all(c("resultado","bancos_privatizados") %in% names(df))) {
  print(t.test(resultado ~ bancos_privatizados, data = df))
}

## 
##  Welch Two Sample t-test
## 
## data:  resultado by bancos_privatizados
## t = 0.81344, df = 49.086, p-value = 0.4199
## alternative hypothesis: true difference in means between group No and group Si is not equal to 0
## 95 percent confidence interval:
##  -1871.452  4417.008
## sample estimates:
## mean in group No mean in group Si 
##         689.7137        -583.0641

# IC para la media de 'resultado' en 2019
if (all(c("anio","resultado") %in% names(df))) {
  x2019 <- df %>% dplyr::filter(anio == 2019) %>% dplyr::pull(resultado)
  if (sum(!is.na(x2019)) > 1) print(t.test(x2019, conf.level = 0.95))
}

# IC para proporción de reestructuración == "Si"
if ("reestructuracion" %in% names(df)) {
  y <- as.integer(df$reestructuracion == "Si")
  print(prop.test(sum(y, na.rm = TRUE), length(y)))
}

## 
##  1-sample proportions test with continuity correction
## 
## data:  sum(y, na.rm = TRUE) out of length(y), null probability 0.5
## X-squared = 23.01, df = 1, p-value = 1.611e-06
## alternative hypothesis: true p is not equal to 0.5
## 95 percent confidence interval:
##  0.6493381 0.8302969
## sample estimates:
##    p 
## 0.75

0.7 6) Apéndice: Top-10 provincias por cantidad de observaciones

if ("provincias" %in% names(df)) {
  df %>%
    count(provincias, sort = TRUE) %>%
    slice_head(n = 10)
}