RPubs link information

• Rpubs link comes here: http://rpubs.com/vaishnavi_reddy/

Introduction

• Context. This project explores a large dataset of recipes to understand how complex different recipes are in terms of preparation and ingredients used.
  
• Focus. The study focuses on measuring complexity through two main factors — the total number of ingredients used and the total number of preparation steps listed for each recipe.
  
• Goal. The main objective is to examine whether recipes that require more ingredients also involve a greater number of steps, by analysing patterns, visualising results, and performing statistical testing to identify any strong relationships.

Introduction Cont.

• In this presentation, we continue exploring how recipe complexity can be measured and compared using statistical methods.
  
• The main aim is to clearly show how the number of ingredients and preparation steps relate to each other.
  
• Each slide will focus on explaining data summaries, visualisations, and hypothesis testing results in a simple and clear manner.
  
• The goal is to make the findings easy to understand and visually appealing while maintaining a proper academic structure.

Problem Statement

• Research question. Is there a positive linear relationship between the number of ingredients and the number of steps in recipes?
• Plan.
  1. Import and describe the dataset.
  2. Visualize complexity by category.
  3. Fit a simple regression model.
  4. Interpret and discuss limitations.

Data

• Source. 1_Recipe_csv (1).csv (provided).
• Unit of analysis. Each row = one recipe.
• Key fields: recipe_title, category, subcategory, description, ingredients, directions, num_ingredients, num_steps.

# Set your file path if it's saved elsewhere
data_path <- "1_Recipe_csv (1).csv"

# Read CSV (base R to avoid extra dependencies)
recipes <- read.csv(data_path, stringsAsFactors = FALSE)

# Basic checks
dim(recipes)

## [1] 62126     8

head(recipes)

Data Cont.

• Variable types.
  • num_ingredients, num_steps are numeric counts.
  • category, subcategory are categorical.
• Pre-processing. Convert text categories to factors for grouped summaries.

recipes$category    <- factor(recipes$category)
recipes$subcategory <- factor(recipes$subcategory)

sapply(recipes[, c("num_ingredients", "num_steps")], function(x) c(
  class = class(x),
  min   = min(x, na.rm = TRUE),
  mean  = mean(x, na.rm = TRUE),
  median= median(x, na.rm = TRUE),
  max   = max(x, na.rm = TRUE),
  n_na  = sum(is.na(x))
))

##        num_ingredients    num_steps         
## class  "integer"          "integer"         
## min    "1"                "1"               
## mean   "9.01744841129318" "4.66186459775295"
## median "9"                "4"               
## max    "35"               "25"              
## n_na   "0"                "0"

Descriptive Statistics and Visualisation

• Overall distributions of num_ingredients and num_steps.
• Relationship between ingredients and steps.

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

par(mfrow = c(1,1))

plot(num_steps ~ num_ingredients, data = recipes,
     main = "Steps vs Ingredients", xlab = "Num Ingredients", ylab = "Num Steps")
abline(lm(num_steps ~ num_ingredients, data = recipes), lwd = 2)

Descriptive Statistics Cont.

• By category: counts and average complexity.

library(dplyr)

by_cat <- recipes %>%
  group_by(category) %>%
  summarise(
    n_recipes       = dplyr::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, 15), caption = "Top 15 Categories by Count with Average Complexity")

Visualisation: Top Categories

• Bar chart of top 10 categories by recipe count.

top10_cats <- by_cat %>% slice_max(n_recipes, n = 10) %>% pull(category)
top10_dat  <- recipes %>% filter(category %in% top10_cats)

bar_dat <- top10_dat %>%
  count(category) %>%
  arrange(desc(n))

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

Visualisation: Complexity by Category

• Boxplots of num_steps by top 8 categories (more readable).

top8_cats <- by_cat %>% slice_max(n_recipes, n = 8) %>% pull(category)
box_dat   <- recipes %>% filter(category %in% top8_cats)

# Base R boxplot to avoid extra dependencies
boxplot(num_steps ~ category, data = box_dat,
        las = 2, cex.axis = 0.8,
        main = "Recipe Steps by Category (Top 8 Categories)",
        xlab = "Category", ylab = "Number of Steps")

Hypothesis Testing

• Model. num_steps = β0 + β1 * num_ingredients + ε
• Hypotheses.
  • \(H_0: beta_1 = 0\) (no linear relationship)
  • \(H_A: beta_1 \ne 0\) (non‑zero linear relationship)

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

## 
## Call:
## lm(formula = num_steps ~ num_ingredients, data = recipes)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -5.0220 -1.5359 -0.2882  1.2164 20.7118 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)     3.545060   0.023203  152.78   <2e-16 ***
## num_ingredients 0.123849   0.002368   52.29   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.261 on 62124 degrees of freedom
## Multiple R-squared:  0.04216,    Adjusted R-squared:  0.04215 
## F-statistic:  2735 on 1 and 62124 DF,  p-value: < 2.2e-16

Hypothesis Testing Cont.

• Optionally adjust for category to account for different cooking styles.
• (Limited to top 10 categories to keep levels manageable in the model output.)

recipes2 <- recipes
recipes2$category_limited <- ifelse(recipes2$category %in% top10_cats,
                                    as.character(recipes2$category), "Other")
recipes2$category_limited <- factor(recipes2$category_limited)

model2 <- lm(num_steps ~ num_ingredients + category_limited, data = recipes2)
summary(model2)

## 
## Call:
## lm(formula = num_steps ~ num_ingredients + category_limited, 
##     data = recipes2)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -5.0527 -1.5190 -0.2844  1.1991 20.8328 
## 
## Coefficients:
##                                      Estimate Std. Error t value Pr(>|t|)    
## (Intercept)                           3.27169    0.05236  62.484  < 2e-16 ***
## num_ingredients                       0.11729    0.00237  49.478  < 2e-16 ***
## category_limitedBeef Recipes          0.61869    0.07739   7.995 1.32e-15 ***
## category_limitedBreads                1.37362    0.07800  17.611  < 2e-16 ***
## category_limitedBreakfast And Brunch  0.04332    0.08040   0.539  0.59003    
## category_limitedCakes                 1.21882    0.07040  17.312  < 2e-16 ***
## category_limitedCookies               0.90579    0.07134  12.696  < 2e-16 ***
## category_limitedDesserts              1.23527    0.07913  15.611  < 2e-16 ***
## category_limitedHealthy Recipes      -0.17449    0.06796  -2.568  0.01024 *  
## category_limitedMain Dishes           1.00269    0.06243  16.062  < 2e-16 ***
## category_limitedOther                 0.19176    0.05010   3.828  0.00013 ***
## category_limitedPork                  0.63541    0.08101   7.843 4.46e-15 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.232 on 62114 degrees of freedom
## Multiple R-squared:  0.06673,    Adjusted R-squared:  0.06657 
## F-statistic: 403.8 on 11 and 62114 DF,  p-value: < 2.2e-16

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: \[S = \sum^n_{i = 1}d^2_i\]

• Findings. The results show that recipes with more ingredients usually require a higher number of preparation steps. This indicates a clear positive relationship between the number of ingredients and the level of complexity in a recipe.
  

• Strengths. The dataset used in this study is large and diverse, which helps provide reliable and meaningful results. The use of simple statistical methods makes the findings easy to interpret and reproduce.
  

• Limitations. The analysis only considers the number of ingredients and steps as measures of complexity. It does not take into account other factors such as cooking time or difficulty of techniques. Some recipe categories may also have uneven data distribution.
  

• Future Work. Future research could include more detailed variables such as preparation time, ingredient types, or cooking methods to better explain recipe complexity and improve the accuracy of predictions.

Conclusion (Take‑Home Message)

• Take-home. The analysis clearly shows that recipes with a higher number of ingredients generally require more preparation steps. This confirms that the ingredient count can be used as a strong indicator of recipe complexity. The findings support the idea that as recipes become more detailed and include more items, the preparation process also becomes longer and more involved.

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).