lgt.R

# ==================================================================
# UNIVERSIDAD CENTRAL DEL ECUADOR
# ANÁLISIS COMPLETO DE LA VARIABLE LONGITUD
# Sedimentos Marinos (2000-2024)
# Autor: Grace Vega - Grupo 2
# ==================================================================

# Ruta correcta (sin comillas curvas)
setwd("C:/Users/Grace/Favorites/Restudio (Estadistica)")

library(dplyr)

## 
## Adjuntando el paquete: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library(knitr)

# 1. Leer y limpiar datos
datos <- read.csv("Sedimentos Marinos.csv", header = TRUE, sep = ";", dec = ".")
longitud <- as.numeric(gsub("[^0-9.-]", "", datos$LONGITUDE))

## Warning: NAs introducidos por coerción

longitud <- na.omit(longitud)

# 2. Cálculos básicos (regla de Sturges)
n  <- length(longitud)
k  <- floor(1 + 3.3 * log10(n))
minimo <- min(longitud)
maximo <- max(longitud)
A  <- (maximo - minimo) / k

# 3. Límites y marca de clase
Li <- seq(minimo, maximo - A, by = A)
Ls <- seq(minimo + A, maximo + 1e-6, by = A)
MC <- (Li + Ls) / 2

# 4. Frecuencias
ni <- numeric(length(Li))
for (i in 1:length(Li)) {
  if (i == length(Li)) {
    ni[i] <- sum(longitud >= Li[i])
  } else {
    ni[i] <- sum(longitud >= Li[i] & longitud < Ls[i])
  }
}

hi    <- round((ni / n) * 100, 4)
Niasc <- cumsum(ni)
Nidsc <- rev(cumsum(rev(ni)))
Hiasc <- round(cumsum(hi), 4)
Hidsc <- round(rev(cumsum(rev(hi))), 4)

# 5. Tabla completa
TDF <- data.frame(
  Li = round(Li, 6), 
  Ls = round(Ls, 6), 
  MC = round(MC, 6),
  ni, hi, Niasc, Nidsc, Hiasc, Hidsc
)

total <- data.frame(
  Li = "--", Ls = "--", MC = "--",
  ni = sum(ni), hi = 100,
  Niasc = "--", Nidsc = "--",
  Hiasc = "--", Hidsc = "--"
)

TDF <- rbind(TDF, total)

# 6. Tabla kable
cat("\n\n====================================================================\n")

## 
## 
## ====================================================================

cat("          TABLA Nº 1 - DISTRIBUCIÓN DE FRECUENCIAS DE LONGITUD        \n")

##           TABLA Nº 1 - DISTRIBUCIÓN DE FRECUENCIAS DE LONGITUD

cat("====================================================================\n")

## ====================================================================

cat("en sedimentos marinos (2000-2024)  |  Autor: Grace Vega - Grupo 2\n\n")

## en sedimentos marinos (2000-2024)  |  Autor: Grace Vega - Grupo 2

kable(
  TDF,
  col.names = c("Lím. Inf.", "Lím. Sup.", "Marca Clase", "ni", "hi (%)",
                "Ni ↑", "Ni ↓", "Hi ↑ (%)", "Hi ↓ (%)"),
  align = "c"
)

Lím. Inf.	Lím. Sup.	Marca Clase	ni	hi (%)	Ni ↑	Ni ↓	Hi ↑ (%)	Hi ↓ (%)
-939.165	-789.423833	-864.294417	261	11.5794	261	2254	11.5794	99.9999
-789.423833	-639.682667	-714.55325	1300	57.6752	1561	1993	69.2546	88.4205
-639.682667	-489.9415	-564.812083	0	0.0000	1561	693	69.2546	30.7453
-489.9415	-340.200333	-415.070917	19	0.8429	1580	693	70.0975	30.7453
-340.200333	-190.459167	-265.32975	4	0.1775	1584	674	70.275	29.9024
-190.459167	-40.718	-115.588583	512	22.7152	2096	670	92.9902	29.7249
-40.718	109.023167	34.152583	36	1.5972	2132	158	94.5874	7.0097
109.023167	258.764333	183.89375	0	0.0000	2132	122	94.5874	5.4125
258.764333	408.5055	333.634917	25	1.1091	2157	122	95.6965	5.4125
408.5055	558.246667	483.376083	77	3.4161	2234	97	99.1126	4.3034
558.246667	707.987833	633.11725	0	0.0000	2234	20	99.1126	0.8873
707.987833	857.729	782.858417	20	0.8873	2254	20	99.9999	0.8873
–	–	–	2254	100.0000	–	–	–	–

# 7. Gráficos
colores <- gray.colors(length(ni), start = 0.3, end = 0.9)

# Histogramas
hist(
  longitud,
  breaks = seq(minimo, maximo + 1e-6, A),
  right = FALSE,
  col = colores,
  main = "Gráfica 1 y 2 - Distribución absoluta de Longitud",
  xlab = "Longitud (grados)",
  ylab = "Frecuencia absoluta"
)

# Barplot
barplot(
  hi[-nrow(TDF)],
  space = 0,
  col = colores,
  names.arg = round(MC, 4),
  main = "Gráfica 3 y 4 - Distribución relativa de Longitud",
  xlab = "Longitud (grados)",
  ylab = "Porcentaje (%)",
  ylim = c(0, 100)
)

# Boxplot corregido
boxplot(
  longitud,
  horizontal = TRUE,
  col = "lightblue",
  main = "Gráfica 5 - Boxplot de Longitud",
  xlab = "Longitud (grados)"
)