Recipe Complexity Analysis

Exploring relationships in a large recipe dataset

VAISHNAVI REDDY TUDI (s4196984)

19 October, 2025

RPubs link information

-RPubs link : http://rpubs.com/vaishnavi_reddy/

Introduction

Context: This study looks at a collection of online recipes to understand what makes some recipes more complex than others. Complexity here means how long or detailed a recipe is to prepare.
Focus: Two simple factors are used to measure complexity — the number of ingredients a recipe has and the number of preparation steps it includes.
Goal: The aim is to find out whether recipes with more ingredients also tend to have more steps. In other words, does adding more ingredients make a recipe harder to prepare?

Introduction Cont:

This presentation explores how the number of ingredients and preparation steps are connected by using simple statistics and graphs.
It demonstrates how basic regression analysis can be used to find patterns and relationships in real data.
The slides are designed to be short, clear, and easy to follow while still showing proper academic structure. .

Problem Statement

Research Question: Does a recipe with more ingredients usually have more preparation steps?
This study aims to find out if there is a clear and positive link between the number of ingredients used and the number of steps involved in cooking a recipe.

Plan:

I will import the recipe dataset and look at the data to understand what it contains.
I will describe the main variables like number of ingredients and number of steps.
I will use summary statistics and graphs to explore how the variables are related.
I will run a regression model to test if more ingredients mean more steps in a recipe.

Data

-Source: 1_Recipe_csv (1).csv (provided dataset).

-Observation: Each row represents one recipe.

Key variables:

-num_ingredients – number of ingredients.

-num_steps – number of preparation steps.

-category – recipe category.

#--- Load Packages and Read Dataset ---



if(!require(janitor)) install.packages("janitor")
if(!require(dplyr)) install.packages("dplyr")
if(!require(knitr)) install.packages("knitr")




library(janitor)
library(dplyr)
library(knitr)

#Read dataset



recipes <- read.csv("1_Recipe_csv (1).csv", stringsAsFactors = FALSE)

#Clean column names



recipes <- janitor::clean_names(recipes)

#--- Dataset Dimensions ---



dims <- data.frame(Rows = nrow(recipes), Columns = ncol(recipes))
knitr::kable(dims, caption = "Dataset Dimensions")

Dataset Dimensions
Rows	Columns
62126	8

#--- Sample of Dataset (Compact View) ---



shorten <- function(x, n = 45){
ifelse(is.na(x), NA, ifelse(nchar(x) > n, paste0(substr(x, 1, n-3), "..."), x))
}




key_cols <- intersect(c("recipe_title","category","subcategory","num_ingredients","num_steps"),
names(recipes))
if(length(key_cols) == 0) key_cols <- names(recipes)[1(5, ncol(recipes))]




preview <- head(recipes[, key_cols, drop = FALSE], 6)
for(col in key_cols){
if(is.character(preview[[col]])) preview[[col]] <- shorten(preview[[col]], 45)
}
knitr::kable(preview, caption = "Sample of Dataset (Compact View)")

Sample of Dataset (Compact View)
recipe_title	category	subcategory	num_ingredients	num_steps
Air Fryer Potato Slices with Dipping Sauce	Air Fryer Recipes	Air Fryer Recipes	9	5
Gochujang Pork Belly Bites	Air Fryer Recipes	Air Fryer Recipes	5	4
3-Ingredient Air Fryer Everything Bagel Ch…	Air Fryer Recipes	Air Fryer Recipes	3	4
Air Fryer Everything Bagel Chicken Cutlets	Air Fryer Recipes	Air Fryer Recipes	9	9
Air Fryer Honey Sriracha Salmon Bites	Air Fryer Recipes	Air Fryer Recipes	5	5
Air Fryer Corn on The Cob	Air Fryer Recipes	Air Fryer Recipes	6	4

#--- Column Type Summary and Numeric Summary ---
#Column Type Summary



col_summary <- data.frame(
Column = names(recipes),
Type = sapply(recipes, function(x) class(x)[1])
)
knitr::kable(col_summary, caption = "Column Type Summary (Clean Output)")

Column Type Summary (Clean Output)
	Column	Type
recipe_title	recipe_title	character
category	category	character
subcategory	subcategory	character
description	description	character
ingredients	ingredients	character
directions	directions	character
num_ingredients	num_ingredients	integer
num_steps	num_steps	integer

#Numeric Summary



numeric_cols <- names(recipes)[sapply(recipes, is.numeric)]




if(length(numeric_cols) > 0){
num_summary <- data.frame(
Variable = numeric_cols,
Min = unname(sapply(recipes[numeric_cols], function(x) min(x, na.rm = TRUE))),
Mean = unname(sapply(recipes[numeric_cols], function(x) mean(x, na.rm = TRUE))),
Median = unname(sapply(recipes[numeric_cols], function(x) median(x, na.rm = TRUE))),
Max = unname(sapply(recipes[numeric_cols], function(x) max(x, na.rm = TRUE))),
Missing = unname(sapply(recipes[numeric_cols], function(x) sum(is.na(x))))
)
num_summary$Mean <- round(num_summary$Mean, 2)
num_summary$Median <- round(num_summary$Median, 2)
knitr::kable(num_summary, caption = "Numeric Summary of Key Variables")
} else {
cat("No numeric variables found in this dataset.")
}

Numeric Summary of Key Variables
Variable	Min	Mean	Median	Max	Missing
num_ingredients	1	9.02	9	35	0
num_steps	1	4.66	4	25	0

Data Cont:

#**The dataset includes some main variables that describe each recipe.

1)recipe_title: The name of the recipe.

2)category: The main group a recipe belongs to, such as “Air Fryer”, “Desserts”, or “Breakfast”.

3)subcategory: A smaller group under each category that gives more detail about the recipe type.

4)num_ingredients: Shows how many ingredients are used in a recipe.

5)num_steps: Shows how many preparation steps are needed to complete the recipe.

-The columns category and subcategory were changed into factors, because they represent types or groups, not numbers. Each level of these factors stands for a different kind of recipe.

-The columns num_ingredients and num_steps were kept as numbers, because they show counts that can be used in calculations like averages or totals.

#**Before starting the analysis, a few cleaning steps were done to make the data ready:

1)Fixed column names so they are easier to use.

2)Changed data types to the correct formats (for example, text to factor, numbers to numeric).

3)Removed rows with missing or wrong values.

4)Checked that all numeric values are correct and make sense.

-These cleaning steps helped make sure the dataset is tidy, accurate, and ready to be used for graphs, summaries, and statistical tests.

Descriptive Statistics and Visualisation

-These plots show the overall distributions of ingredients and steps.

-A scatterplot is used to observe any visible relationship.

par(mfrow = c(1,2))
hist(recipes$num_ingredients, main = "Number of Ingredients", xlab = "Ingredients", col = "lightblue")
hist(recipes$num_steps, main = "Number of Steps", xlab = "Steps", col = "lightgreen")

par(mfrow = c(1,1))




plot(recipes$num_ingredients, recipes$num_steps,
main = "Steps vs Ingredients",
xlab = "Number of Ingredients",
ylab = "Number of Steps",
pch = 19, col = "blue")
abline(lm(num_steps ~ num_ingredients, data = recipes), col = "red", lwd = 2)

Descriptive Statistics Cont.

-Summary by category showing average ingredients and steps.

by_cat <- recipes %>%
group_by(category) %>%
summarise(
n_recipes = n(),
avg_ingredients = mean(num_ingredients, na.rm = TRUE),
avg_steps = mean(num_steps, na.rm = TRUE)
) %>%
arrange(desc(n_recipes))




knitr::kable(head(by_cat, 10), caption = "Top 10 Categories by Count and Average Complexity")

Top 10 Categories by Count and Average Complexity
category	n_recipes	avg_ingredients	avg_steps
Main Dishes	3387	10.421022	5.496605
Healthy Recipes	2237	7.724184	4.003129
Appetizers And Snacks	2084	7.903551	4.198656
Cakes	1954	9.615148	5.618219
Cookies	1849	8.843699	5.214711
Beef Recipes	1400	10.569286	5.130000
Breads	1352	9.116124	5.714497
Desserts	1288	7.242236	5.356367
Breakfast And Brunch	1223	8.147997	4.270646
Pork	1205	10.507054	5.139419

Visualisation: Top Categories

-Bar chart of the top 10 categories by number of recipes.

#Make sure data is available



if (!exists("recipes")) {
recipes <- read.csv("1_Recipe_csv (1).csv", stringsAsFactors = FALSE)
}

#Only proceed if 'category' exists



if ("category" %in% names(recipes)) {

#Count recipes per category and take top 10



library(dplyr)
top10 <- recipes %>%
count(category, name = "n_recipes") %>%
arrange(desc(n_recipes)) %>%
slice_head(n = 10)

#Draw bar plot



barplot(height = top10$n_recipes,
names.arg = as.character(top10$category),
las = 2, cex.names = 0.8,
main = "Top 10 Categories (Count of Recipes)",
ylab = "Count")




} else {
cat("Note: 'category' column not found in the dataset; cannot plot top categories.\n")
}

Hypothesis Testing

Model: num_steps = β0 + β1 * num_ingredients + ε
Hypotheses:
- \(H_0: \beta_1 = 0\) (No linear relationship between number of ingredients and steps)
- \(H_A: \beta_1 \ne 0\) (A significant linear relationship exists)

# Safety check: reload data if missing
if (!exists("recipes")) {
  recipes <- read.csv("1_Recipe_csv (1).csv", stringsAsFactors = FALSE)
}
#Simple Linear Regression Model



model1 <- lm(num_steps ~ num_ingredients, data = recipes)

#Show model code and clean table result


summary(model1)$coefficients["num_ingredients", , drop = FALSE] |>
as.data.frame() |>
tibble::rownames_to_column("Variable") |>
knitr::kable(
caption = "T-Test for num_ingredients in Simple Linear Regression"
)

T-Test for num_ingredients in Simple Linear Regression
Variable	Estimate	Std. Error	t value	Pr(>\|t\|)
num_ingredients	0.1238493	0.0023683	52.29418	0

#Show key model metrics neatly


summary_table <- data.frame(
R_Squared = round(summary(model1)$r.squared, 4),
Adj_R_Squared = round(summary(model1)$adj.r.squared, 4),
F_Statistic = round(summary(model1)$fstatistic[1], 3),
P_Value = signif(pf(summary(model1)$fstatistic[1],
summary(model1)$fstatistic[2],
summary(model1)$fstatistic[3],
lower.tail = FALSE), 4)
)
knitr::kable(summary_table, caption = "Model Summary Statistics")

Model Summary Statistics
	R_Squared	Adj_R_Squared	F_Statistic	P_Value
value	0.0422	0.0421	2734.681	0

Hypothesis Testing Cont.

The model is adjusted for recipe category (top 10 + “Other”) to check if cooking styles affect the relationship.
This helps control for different cooking styles.

#Adjust for top 10 categories



top10_cats <- names(sort(table(recipes$category), decreasing = TRUE))[1:10]
recipes$category_limited <- ifelse(recipes$category %in% top10_cats, recipes$category, "Other")
recipes$category_limited <- factor(recipes$category_limited)

#Multiple Linear Regression



model2 <- lm(num_steps ~ num_ingredients + category_limited, data = recipes)

#Clean coefficient table



summary(model2)$coefficients["num_ingredients", , drop = FALSE] |>
as.data.frame() |>
tibble::rownames_to_column("Variable") |>
knitr::kable(
caption = "T-Test for num_ingredients (Adjusted for Category)"
)

T-Test for num_ingredients (Adjusted for Category)
Variable	Estimate	Std. Error	t value	Pr(>\|t\|)
num_ingredients	0.1172847	0.0023704	49.47811	0

#Add R-squared and F-stat summary for adjusted model


summary_table2 <- data.frame(
R_Squared = round(summary(model2)$r.squared, 4),
Adj_R_Squared = round(summary(model2)$adj.r.squared, 4),
F_Statistic = round(summary(model2)$fstatistic[1], 3),
P_Value = signif(pf(summary(model2)$fstatistic[1],
summary(model2)$fstatistic[2],
summary(model2)$fstatistic[3],
lower.tail = FALSE), 4)
)
knitr::kable(summary_table2, caption = "Adjusted Model Summary Statistics")

Adjusted Model Summary Statistics
	R_Squared	Adj_R_Squared	F_Statistic	P_Value
value	0.0667	0.0666	403.778	0

Mathematical Equations (Examples)

Inline: \(z = \frac{x - \mu}{\sigma}\)
Two-sample means: \[H_0: \mu_1 = \mu_2 \] -\[H_A: \mu_1 \ne \mu_2\]
Sum of squares: \[S = \sum^n_{i = 1}d^2_i\]

#Discussion

Findings: The results show that recipes with more ingredients usually have more preparation steps. This means there is a clear positive connection between the two variables.
Strengths: The dataset has many recipes from different categories, which helps make the results more reliable. The method used is simple and easy to understand.
Limitations: The study only looks at two things — ingredients and steps. It does not include cooking time, skill level, or other details that could affect complexity.
Future Work: In the future, more factors like cooking time or ingredient type can be added to make the analysis more accurate and detailed.

Conclusion (Take-Home Message)

Recipes that have more ingredients usually take more steps to prepare.
This shows that the number of ingredients is a good way to understand how complex a recipe is.
Even when recipe categories are considered, this pattern stays the same.

References

R Core Team (2025). R: A Language and Environment for Statistical Computing.
Wickham, H. (2024). dplyr and ggplot2 R packages.
Class notes and lecture materials (2025).