Dumaguete City Temperature – Descriptive Statistics

Raw Data (Hourly Temperature, °C)

Computed Descriptive Statistics

R Code to Compute Descriptive Statistics

# START

#=============================================
# Filename: descriptive_stats.R
# Purpose: Small-sample descriptive statistics
#=============================================

#------------------------------
# Load necessary packages
#------------------------------
library(dplyr)   # For data cleaning, filtering, and piping (%>%)
library(readr)   # For fast, consistent reading of rectangular data (like CSVs)
library(e1071)   # For moments (skew/kurtosis) and various machine learning tools

#------------------------------
# Read CSV data
#------------------------------
# Place your CSV file in the working directory or specify full path
temp_data <- read_csv("2026-01-01_DUM.csv")

# View first few rows to confirm
print(head(temp_data))

#------------------------------
# Extract temperature vector
#------------------------------
temps <- temp_data$t2m_C  # column with hourly temperatures

#------------------------------
# Basic statistics
#------------------------------
n        <- length(temps)                     # sample size
mean_val <- mean(temps)                       # mean
max_val  <- max(temps)                        # maximum
min_val  <- min(temps)                        # minimum
median_val <- median(temps)                   # median
range_val  <- max_val - min_val               # range
sd_val     <- sd(temps)                       # sample standard deviation
var_val    <- var(temps)                      # sample variance
mad_val    <- mean(abs(temps - mean_val))     # mean absolute deviation

#------------------------------
# Percentiles (inclusive, small sample)
#------------------------------
q1  <- quantile(temps, 0.25, type = 7)   # 25th percentile (inclusive)
q3  <- quantile(temps, 0.75, type = 7)   # 75th percentile (inclusive)
iqr_val <- IQR(temps)                     # interquartile range
p90 <- quantile(temps, 0.90, type = 7)   # 90th percentile
p99 <- quantile(temps, 0.99, type = 7)   # 99th percentile

#------------------------------
# Skewness and Kurtosis
#------------------------------
# Using e1071 package:
# type = 2: Bias-corrected for small samples.
# This matches Excel/LibreOffice SKEW() and KURT() functions.
# For small samples (n < 50), this provides a more accurate estimate of the population.

skew <- e1071::skewness(temps, type = 2)  # Matches Excel SKEW() (Symmetry: 0 = symmetric)
kurt <- e1071::kurtosis(temps, type = 2)  # Matches Excel KURT() (Excess Kurtosis: 0 = normal)

#------------------------------
# Coefficient of Variation
#------------------------------
cv <- sd_val / mean_val  # SD standardized by mean

#------------------------------
# Z-scores
#------------------------------
z_scores <- (temps - mean_val) / sd_val

#------------------------------
# Summary table
#------------------------------
summary_table <- data.frame(
  Measure = c(
    "Sample Size", "Mean (°C)", "Maximum (°C)", "Minimum (°C)",
    "Median (°C)", "Range (°C)", "Sample SD (°C)", "Variance (°C²)",
    "Mean Absolute Deviation (°C)", "25th inclusive percentile, Q1 (°C)",
    "75th inclusive percentile, Q3 (°C)", "Interquartile Range (°C)",
    "90th percentile (°C)", "99th percentile (°C)", "Skewness (type=2, e1071)",
    "Kurtosis (Fisher, type=2, e1071)", "Coefficient of Variation (CV)"
  ),
  Value = c(
    n, round(mean_val,2), round(max_val,2), round(min_val,2),
    round(median_val,2), round(range_val,2), round(sd_val,2), round(var_val,2),
    round(mad_val,2), round(q1,2), round(q3,2),
    round(iqr_val,2), round(p90,2), round(p99,2),
    round(skew,2), round(kurt,2), round(cv,3)
  )
)

#------------------------------
# Print summary table
#------------------------------
cat("\n=== Descriptive Statistics Summary ===\n")
print(summary_table)

#------------------------------
# Print Full 24-Hour Z-score Report
#------------------------------
cat("\n=== 24-Hour Temperature Stability Report ===\n")

# Create the full table using all rows (n)
z_table_full <- data.frame(
  "Hour"    = temp_data$valid_time,
  "Temp_C"  = temp_data$t2m_C,
  "Z_score" = round(as.numeric(z_scores), 2)
)

# Print the entire table
# We use print.data.frame to ensure all 24 rows show up in the console
print(z_table_full, row.names = FALSE)

cat("\n(Analysis complete for all", nrow(temp_data), "hours)\n")

# END

RStudio Workflow to Compute Descriptive Statistics

1. Ensure the "stats" project folder exists.
2. Installs these additional packages at once: e1071 (for skewness and kurtosis), dplyr (for data manipulation), and readr (for reading CSV files), once per project:
>

install.packages(c("e1071", "dplyr", "readr"))

3. Download the CSV file (2026-01-01_DUM.csv) and place it inside the "stats" folder.
4. Create a new R script in RStudio:
   • Click File → New File → R Script
   • Save the script as descriptive_stats.R inside the "stats" folder
5. Copy the R code to compute descriptive statistics into descriptive_stats.R.
6. Save the script and click Source (top right of the script editor) to run all commands, or issue the command:
>

source("descriptive_stats.R")

7. Check the Console for the printed table of descriptive statistics.
8. Optional: Snapshot the environment for reproducibility:
   >

   renv::snapshot()

The Significance of Z-Scores

It is important to understand the Z-score. A Z-score tells us how far a specific temperature is from the daily average in terms of standard deviations.

Raw Hourly Data & Z-Score Report

Summary of Key Descriptive Measures

Statistical Validation Files

To confirm the statistical accuracy of the R script, you can compare the outputs against these pre-configured spreadsheets. Both files contain the same data and calculations.

• Option A (Excel): Download 2026-01-01_DUM.xlsx
• Option B (LibreOffice): Download 2026-01-01_DUM.ods

How to Verify:

1. Open your preferred file in Microsoft Excel or LibreOffice Calc.
2. Locate the Descriptive Statistics and Shape Analysis sections.
3. Verify that the =SKEW() and =KURT() results match the R script's type = 2 outputs exactly.

Note: LibreOffice is a free, open-source alternative available at the official download site.

References

• Green, J. L., Manski, S. E., Hansen, T. A., & Broatch, J. E. (2022). Descriptive statistics. In Elsevier eBooks (pp. 723–733). https://doi.org/10.1016/b978-0-12-818630-5.10083-1
• Hersbach, H., Bell, B., Berrisford, P., Biavati, G., Horányi, A., Muñoz Sabater, J., Nicolas, J., Peubey, C., Radu, R., Rozum, I., Schepers, D., Simmons, A., Soci, C., Dee, D., & Thépaut, J.-N. (2023). ERA5 hourly data on single levels from 1940 to present [Data set]. Copernicus Climate Data Store (CDS). https://doi.org/10.24381/cds.adbb2d47
• Navarro, D., & Foxcroft, D. (2025). 4. Descriptive statistics. In Open Book Publishers (pp. 59–80). https://doi.org/10.11647/obp.0333.04