20 casos de uso prácticos para trabajar con datos en R: limpieza, análisis exploratorio, segmentación, visualización, pruebas estadísticas, modelos simples y reportes reproducibles.
20 practical R use cases for data work: cleaning, exploratory analysis, segmentation, visualization, statistical testing, simple models, and reproducible reporting.
Si PowerShell es fuerte para automatizar tareas operativas y Python es excelente para integrar datos desde múltiples fuentes, R se destaca como una herramienta poderosa para entender el comportamiento de los datos: tendencias, distribuciones, patrones, grupos, relaciones y señales estadísticas.
R no es solamente para académicos. En un ambiente de BI, gobierno, educación, salud, recursos humanos o finanzas, puede servir como una capa analítica para validar hipótesis, preparar visualizaciones, crear modelos simples y documentar análisis de forma reproducible.
If PowerShell is strong for operational automation and Python is excellent for integrating data from multiple sources, R stands out as a powerful tool for understanding data behavior: trends, distributions, patterns, groups, relationships, and statistical signals.
R is not only for academics. In BI, government, education, healthcare, human resources, or finance environments, it can serve as an analytical layer to validate hypotheses, prepare visualizations, create simple models, and document analysis in a reproducible way.
Antes de limpiar o analizar, R permite revisar tipos de datos, valores faltantes y forma general del dataset.
Before cleaning or analyzing, R helps inspect data types, missing values, and the overall shape of the dataset.
library(readr)
library(dplyr)
data <- read_csv("survey_results.csv")
glimpse(data)
summary(data)Estandariza campos como departamento, programa o modalidad para evitar categorías duplicadas.
Standardize fields such as department, program, or modality to avoid duplicate categories.
library(stringr)
data <- data %>%
mutate(
department_clean = department %>%
str_squish() %>%
str_to_title()
)Sirve para saber qué variables son confiables y cuáles necesitan revisión antes de un dashboard.
Useful for knowing which variables are reliable and which ones need review before a dashboard.
missing_summary <- data %>%
summarise(across(
everything(),
~ mean(is.na(.)) * 100
))
missing_summaryDetecta duplicados usando una combinación de campos, no solo un ID.
Detect duplicates using a combination of fields, not just one ID.
duplicates <- data %>%
group_by(first_name, last_name, date_of_birth) %>%
filter(n() > 1) %>%
arrange(last_name, first_name)
duplicatesConvierte una fecha en variables útiles para tendencias y análisis temporal.
Turn one date into useful variables for trend and time-based analysis.
library(lubridate)
data <- data %>%
mutate(
date_clean = mdy(transaction_date),
year = year(date_clean),
month = month(date_clean, label = TRUE),
quarter = quarter(date_clean)
)Resume métricas por departamento, campus, programa o cualquier grupo administrativo.
Summarize metrics by department, campus, program, or any administrative group.
summary_by_dept <- data %>%
group_by(department_clean) %>%
summarise(
records = n(),
avg_score = mean(score, na.rm = TRUE),
median_score = median(score, na.rm = TRUE)
) %>%
arrange(desc(records))Ayuda a ver si una métrica está concentrada, dispersa, sesgada o con valores extremos.
Helps reveal whether a metric is concentrated, dispersed, skewed, or affected by outliers.
library(ggplot2)
ggplot(data, aes(x = score)) +
geom_histogram(bins = 30) +
labs(
title = "Score Distribution",
x = "Score",
y = "Records"
)Encuentra valores que se salen del comportamiento típico sin asumir una distribución normal.
Find values outside typical behavior without assuming a normal distribution.
q1 <- quantile(data$amount, 0.25, na.rm = TRUE)
q3 <- quantile(data$amount, 0.75, na.rm = TRUE)
iqr <- q3 - q1
outliers <- data %>%
filter(amount < q1 - 1.5 * iqr | amount > q3 + 1.5 * iqr)Combina una tabla de respuestas con una tabla maestra de empleados, estudiantes o cuentas.
Combine a response table with a master table of employees, students, or accounts.
master <- read_csv("employee_master.csv")
combined <- data %>%
left_join(master, by = "employee_id")Detecta empleados, estudiantes o transacciones que aparecen en una tabla pero no en la maestra.
Detect employees, students, or transactions that appear in one table but not in the master table.
not_found <- data %>%
anti_join(master, by = "employee_id")
not_foundGenera una matriz resumida similar a Excel, útil para validar antes de Power BI.
Generate an Excel-like summary matrix, useful for validation before Power BI.
pivot_table <- data %>%
count(department_clean, status) %>%
tidyr::pivot_wider(
names_from = status,
values_from = n,
values_fill = 0
)Transforma encuestas de opinión en valores numéricos para promedios y comparaciones.
Transform opinion surveys into numeric values for averages and comparisons.
likert_map <- c(
"Strongly Disagree" = 1,
"Disagree" = 2,
"Neutral" = 3,
"Agree" = 4,
"Strongly Agree" = 5
)
data <- data %>%
mutate(q1_score = recode(q1_response, !!!likert_map))Crea una primera lectura de temas repetidos en respuestas abiertas.
Create a first reading of repeated themes in open-ended responses.
library(tidytext)
word_counts <- data %>%
unnest_tokens(word, comment) %>%
anti_join(stop_words, by = "word") %>%
count(word, sort = TRUE)
head(word_counts, 20)Evalúa si dos métricas se mueven juntas, por ejemplo satisfacción y tiempo de respuesta.
Evaluate whether two metrics move together, such as satisfaction and response time.
correlation <- cor(
data$satisfaction_score,
data$response_time,
use = "complete.obs"
)
correlationPermite probar si la diferencia observada entre dos grupos puede ser significativa.
Test whether the observed difference between two groups may be statistically meaningful.
test_result <- t.test(
score ~ training_received,
data = data
)
test_resultModela una variable continua usando factores explicativos simples.
Model a continuous variable using simple explanatory factors.
model <- lm(
satisfaction_score ~ response_time + training_hours,
data = data
)
summary(model)Agrupa registros parecidos según variables numéricas para descubrir segmentos.
Group similar records based on numeric variables to discover segments.
features <- data %>%
select(score, response_time, training_hours) %>%
na.omit() %>%
scale()
set.seed(123)
clusters <- kmeans(features, centers = 3)
clusters$centersConvierte transacciones diarias en una vista mensual para detectar crecimiento, caída o estacionalidad.
Convert daily transactions into a monthly view to detect growth, decline, or seasonality.
monthly_trend <- data %>%
mutate(month_start = floor_date(date_clean, "month")) %>%
group_by(month_start) %>%
summarise(total_amount = sum(amount, na.rm = TRUE))
ggplot(monthly_trend, aes(month_start, total_amount)) +
geom_line() +
labs(title = "Monthly Trend")Entrega resultados limpios, resúmenes y excepciones en un solo archivo revisable.
Deliver clean results, summaries, and exceptions in one reviewable workbook.
library(writexl)
write_xlsx(
list(
Clean_Data = data,
Missing_Summary = missing_summary,
Duplicates = duplicates,
Outliers = outliers
),
"r_data_quality_report.xlsx"
)Documenta análisis, código, gráficos y conclusiones en un reporte que se puede volver a generar.
Document analysis, code, charts, and conclusions in a report that can be regenerated.
library(rmarkdown)
render(
input = "analysis_report.Rmd",
output_format = "html_document",
output_file = "analysis_report.html"
)