Analysis of Fast Food Calories

INFO 523 - Project 1

Abstract

Introduction

Introducing the dataset

The “Fast Food Calories” dataset comprises nutritional information for various food items offered by eight distinct fast-food outlets. Each outlet’s menu is detailed, encompassing items such as burgers, fries, beverages, and salads. The dataset’s sources include: research institutions, open data reposiotories(tidytuesday) and compiled data sets from various sources. The dataset contains 18 columns representing different attributes of fast-food items, such as name, restaurant, category, calorie count, fat content, protein content, carbohydrate content, etc. The 515 rows corresponds to the number of individual fast-food items for each restaurant in the dataset.

Exploratory Data Analysis

Question 1

What is the median calorie consumption per visit at each fast food restaurant or outlet?

Introduction

This question allows us to grasp the median calorie intake per visit across different fast food restaurants, providing insights into their nutritional profiles. By analysing the typical calorie intake, we can learn a lot about the energy content of foods served in popular fast food outlets. Examining median calorie consumption each visit acts as a realistic nutritional guide for people who are conscious of their meal choices, providing helpful information in choosing healthier eating habits. It can also be an important resource for government bodies concerned with public health, giving information that can be used to develop policies aiming at increasing nutritional standards and fighting diet-related health concerns in communities.

Approach

We’ll use the “groupby.describe()” method to create a summary table with contains statistical information like mean, mode, median, minimum, and maximum calorie content for each item in every restaurant. The variables used for this analysis are Restaurant, Item, Total Fat, and Calories. We computed the median calories consumed per person due to the distributional nature of the data. This data is visualised using a violin plot, which will provide a clear picture of the distribution of median calorie intake per person across different eateries, allowing for a better understanding and interpretation of nutrition patterns. The line plot with triangle marker is used to visualise the median calories of each restuarent. The median which is used as a central tendency metric is best visualised using the line plot. We also have derived summary statistics atble for each restuarent.

Analysis

Part 1

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Load the dataset
df = pd.read_csv("data/fastfood_calories.csv")

# Group by restaurant and calculate the median calories per visit
median_calories_per_visit = df.groupby('restaurant')['calories'].median()

# Print summary statistics
summary_table = df.groupby('restaurant')['calories']

# Sort the DataFrame by restaurant names
df_sorted = df.sort_values(by='restaurant')

# Plotting violin plot
plt.figure(figsize=(12, 8))
violin = sns.violinplot(x='calories', y='restaurant', data=df_sorted, inner='quartile', cut=0, palette='husl')
plt.title('Distribution of Calories per Visit by Restaurant', fontweight='bold')
plt.xlabel('Calories', fontweight='bold')
plt.ylabel('Restaurant', fontweight='bold')

# Create legend using violinplot handles
plt.legend(title='Median Calories', handles=violin.collections, labels=[f"{int(median_calories)} calories" 
                                                                         for median_calories in median_calories_per_visit],
           loc='upper left', bbox_to_anchor=(1, 1), prop={'weight': 'bold'})

plt.show()

print("Summary Table:")
summary_table.describe( )

Summary Table:

	count	mean	std	min	25%	50%	75%	max
restaurant
Arbys	55.0	532.727273	210.338832	70.0	360.0	550.0	690.0	1030.0
Burger King	70.0	608.571429	290.418417	190.0	365.0	555.0	760.0	1550.0
Chick Fil-A	27.0	384.444444	220.494782	70.0	220.0	390.0	480.0	970.0
Dairy Queen	42.0	520.238095	259.337694	20.0	350.0	485.0	630.0	1260.0
Mcdonalds	57.0	640.350877	410.696120	140.0	380.0	540.0	740.0	2430.0
Sonic	53.0	631.698113	300.881627	100.0	410.0	570.0	740.0	1350.0
Subway	96.0	503.020833	282.220965	50.0	287.5	460.0	740.0	1160.0
Taco Bell	115.0	443.652174	184.344883	140.0	320.0	420.0	575.0	880.0

Part 2

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Load the dataset into a DataFrame
calories_df = pd.read_csv('data/fastfood_calories.csv')

# Calculate median calories for each restaurant
median_calories = calories_df.groupby('restaurant')['calories'].median().reset_index()

# Set the color palette
colors = sns.color_palette("husl", len(median_calories['restaurant']))

# Set the figure size
plt.figure(figsize=(10, 6))

# Create the line plot with inverted triangles and white background
for i, restaurant in enumerate(median_calories['restaurant']):
    plt.plot(restaurant, median_calories.loc[i, 'calories'], marker='v', markersize=12, color=colors[i], label=restaurant)
    plt.text(restaurant, median_calories.loc[i, 'calories'] - 15, f"{median_calories.loc[i, 'calories']:.0f}", ha='center', fontsize=10, color='black')

# Add title and labels
plt.title('Average Calories per visit to a Restaurant', fontsize=16)
plt.xlabel('Restaurant', fontsize=14)
plt.ylabel('Average Calories per visit', fontsize=14)

# Rotate x-axis labels for better readability (optional)
plt.xticks(rotation=45, ha='right')

# Set y-axis limit
plt.ylim(350, 650)

# Set y-axis ticks with step 50
plt.yticks(range(350, 651, 50))

# Remove gridlines
plt.grid(False)

Discussion

The distribution of calorie consumption per visit across various fast-food restaurants is illustrated by the visualization of a violin plot. The width of each “violin” in the violin plot, corresponds to the density of observations at various calorie levels. Each violin’s quartiles provide information on the data’s central tendency and distribution, aiding in the understanding of the variability in calorie counts among restaurants. A summary table is printed, which includes information such as mean, standard deviation, minimum, 25th percentile, median (50th percentile), 75th percentile, and maximum calorie counts. This mix of graphic and numerical statistics offers a thorough picture of calorie consumption patterns across different fast-food businesses. The second visualization presents a line plot featuring inverted triangle marker, where each restaurant median calorie is symbolized by a marker. Horizontally positioned according to restaurant names and vertically aligned based on the average calorie count per visit, the plot enables us to discern that certain restaurants boast notably higher median calorie counts. This highlights the substantial variation in the nutritional profiles of meals served at different fast-food establishments. Some restaurants exhibit consistent calorie counts across visits, while others demonstrate greater variability, indicative of a broader selection of food options with varying calorie content.

Question 2

How do different food item categories vary in protien-fat ratio and do they meet up to the standards of the health metric?

Introduction

The goal of this question is to examine the wide range of protein-fat ratios seen in different dietary categories and determine how well they correspond with recognized health measures. Examining the variations in protein-fat proportions among different types of food items is crucial for evaluating their nutritional value. By analyzing these ratios and comparing them with health benchmarks, we can identify food categories that meet recommended nutritional criteria and those that may benefit from adjustments to optimize health benefits. Through this analysis, we aim to gain a deeper understanding of the complex interplay between dietary composition and its impact on health outcomes.

Approach

For question 2, we’ll use the “protein” and “item” variables. To standardize the proteins and fats we’ll convert them to calories. By employing the protein-fat ratio, we established a health metric to classify food items as healthy or unhealthy. We have used a half-violin plot to visualize these findings. The half-violin plot is particularly helpful since it demonstrates the distribution of our data while simultaneously indicating its symmetry. The shape of the plot gives us insight into the density of our data, which helps us comprehend the distribution of our health metrics. We have also used a scatter plot to visualize the median protein-fat ratio of each restaurant. Using these Scatter plots provides a clear visual representation of the data points’ dispersion. The summary statistics table for the menu items offered by each of these eateries was found based on the protein-fat ratio.

Analysis

Part 1

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Load the dataset into a DataFrame
calories_df = pd.read_csv('data/fastfood_calories.csv')

# Define calorie-per-gram values for fat and protein
calories_per_gram_fat = 9  # Fat provides 9 calories per gram
calories_per_gram_protein = 4  # Protein provides 4 calories per gram

# Calculate calories from protein and fat
calories_df['cal_protein'] = calories_df['protein'] * calories_per_gram_protein
calories_df['cal_fat'] = calories_df['total_fat'] * calories_per_gram_fat

# Calculate total calories
calories_df['total_calories'] = calories_df['cal_protein'] + calories_df['cal_fat']

# Calculate Protein_fat_ratio
calories_df['Protein_fat_ratio'] = calories_df['cal_protein'] / calories_df['total_calories']

# Calculate Q2 (median) for each restaurant
median_protein_fat_ratio = calories_df.groupby('restaurant')['Protein_fat_ratio'].median().reset_index()

# Set custom color palette
colors = sns.color_palette("husl", n_colors=len(median_protein_fat_ratio))

# Create a half violin plot to visualize the median Protein_fat_ratio for each restaurant
plt.figure(figsize=(12, 8))
sns.violinplot(x='Protein_fat_ratio', y='restaurant', data=calories_df, inner='quartile', linewidth=1.5, cut=0, split=True, palette=colors)

# Add Q2 values (medians) to the plot without labels
for i in range(len(median_protein_fat_ratio)):
    plt.text(median_protein_fat_ratio['Protein_fat_ratio'][i], i, f"{median_protein_fat_ratio['Protein_fat_ratio'][i]:.2f}", fontsize=10, va='center', ha='right', color='black')

# Add title and labels
plt.title('Protein-Fat Ratio by Restaurant', fontsize=16)
plt.xlabel('Protein-Fat Ratio', fontsize=14)
plt.ylabel('Restaurant', fontsize=14)

# Customize y-axis labels
plt.yticks(fontsize=12)

# Show the plot
plt.grid(True, linestyle='--', alpha=0.7)
plt.tight_layout()
plt.show()

The table below represents the health metric for the fast food restaurants based on the protein fat ratio.

import matplotlib.pyplot as plt

# Define data
protein_fat_ratios = [0.35, 0.26, 0.49, 0.26, 0.33, 0.26, 0.47, 0.26]
restaurants = ['Arbys', 'Burger King', 'Chick Fil-A', 'Dairy Queen', 'McDonalds', 'Sonic', 'Subway', 'Taco Bell']

# Create the plot
plt.figure(figsize=(10, 6))

# Plot points for protein fat ratios for each restaurant
for i, restaurant in enumerate(restaurants):
    plt.scatter([protein_fat_ratios[i]], [restaurant], label=restaurant, marker='o')

# Shade the regions
plt.axvspan(0.25, 0.40, color='lightcoral', alpha=0.5, label='Needs Improvement')
plt.axvspan(0.40, 0.50, color='lightgreen', alpha=0.5, label='Healthy')

# Labels and legend
plt.xlabel('Protein Fat Ratio')
plt.ylabel('Restaurants')
plt.title('Protein Fat Ratio of Restaurants')

# Customize y-axis labels
plt.yticks(restaurants)

# Show plot
plt.grid(True, linestyle='--', alpha=0.7)
plt.tight_layout()
#plt.legend()
plt.show()

The code chunk below represents the summary statistics of the food categories of the data set using the protein fat ratio

import pandas as pd
import numpy as np
calories_df = pd.read_csv("data/fastfood_calories.csv")
#below lies the dictionary for the categorisation of the food items into groups
categories = {
    'Burgers': ['Burger', 'Cheeseburger', 'McNuggets', 'Quarter Pounder'],
    'Chicken Sandwiches': ['Chicken Sandwich', 'Grilled Chicken', 'Crispy Chicken'],
    'Salads': ['Salad'],
    'Chicken Nuggets/Tenders': ['Chicken McNuggets', 'Chicken Tenders'],
    'Hot Dogs': ['Hot Dog'],
    'Subway Items': ['Subway'],
    'Tacos/Burritos': ['Taco', 'Burrito', 'Nachos']
}
##the function below was used to group the food groups based on the presence of certain words in the dictionary
def categorize_item(item):
    for category, keywords in categories.items():
        if any(keyword.lower() in item.lower() for keyword in keywords):
            return category
    return 'Other'
  
calories_df['categorized_food_items'] = calories_df['item'].apply(categorize_item)
###creating the protein fat ratio for the columns in the dataset
# Define calorie-per-gram values for fat and protein
calories_per_gram_fat = 9  # Fat provides 9 calories per gram
calories_per_gram_protein = 4  # Protein provides 4 calories per gram

# Calculate calories from protein and fat
calories_df['cal_protein'] = calories_df['protein'] * calories_per_gram_protein
calories_df['cal_fat'] = calories_df['total_fat'] * calories_per_gram_fat

# Calculate total calories
calories_df['total_calories'] = calories_df['cal_protein'] + calories_df['cal_fat']

# Calculate Protein_fat_ratio
calories_df['Protein_fat_ratio'] = calories_df['cal_protein'] / calories_df['total_calories']


summary_stats = calories_df.groupby('categorized_food_items')['Protein_fat_ratio']

summary_stats.describe( )

	count	mean	std	min	25%	50%	75%	max
categorized_food_items
Burgers	63.0	0.292165	0.042541	0.216471	0.265655	0.287770	0.322581	0.419355
Chicken Nuggets/Tenders	3.0	0.383500	0.008621	0.375510	0.378932	0.382353	0.387495	0.392638
Chicken Sandwiches	81.0	0.378670	0.159336	0.092742	0.267223	0.346604	0.476744	0.836364
Hot Dogs	3.0	0.223662	0.010229	0.213592	0.218471	0.223350	0.228696	0.234043
Other	252.0	0.347719	0.130331	0.123515	0.252846	0.320143	0.400865	0.697987
Salads	43.0	0.403233	0.215495	0.095694	0.258310	0.307692	0.561957	1.000000
Subway Items	6.0	0.462131	0.196619	0.233184	0.295307	0.481675	0.624068	0.671533
Tacos/Burritos	63.0	0.299672	0.075370	0.170213	0.250000	0.289474	0.339153	0.550000

Discussion

The half-violin plot depicts how the protein-fat ratio varies between fast-food outlets. The numbers near quartile lines inside the violin plots show each restaurant’s median protein-fat ratio. The quartile line range provides information on the nutritional markup of their menu items. Some restaurants have narrower distributions, indicating less protein-fat ratios in those quartile distributions, while others have larger distributions, implying more protein-fat ratio in the quartile ranges. The line plot with a circle marker depicts the protein-fat ratios of several fast-food establishments. Each point in the diagram represents a restaurant, with the placement determined by the protein-fat ratio. The shaded regions determine the threshold for classifying restaurants as either “Needs Improvement” or “Healthy” based on their protein-fat ratios. Restaurants classified in the “Needs Improvement” category suggest that their menu options are characterized by a higher proportion of fat relative to protein. Conversely, eateries placed in the “Healthy” category indicate a more favorable protein-to-fat ratio, implying a healthier balance in their menu offerings. For the summary statistics table the food items were categorized into 8 main categories, burgers, chicken nuggets/ tenders, chicken sandwiches, hotdogs, other, salads, subway items, and tacos and burritos, these categorizations were done using a dictionary. A dictionary was created for each food category containing certain words distinct to each food category. Based on the words in the dictionary for each food category, if a word in the food item can be found in the food category dictionary, the food item will be categorized as such.

Group members

• Valerie Naa Dei Okine: First year graduate student Data Science at University of Arizona.

• Sai Navya Reddy: First year graduate student Data Science at University of Arizona.

• Gowtham Theeda: First year graduate student Data Science at University of Arizona.

• Tanya Evita George: First year graduate student Data Science at University of Arizona.

• Monica Tejaswi Kommareddy:First year graduate student Data Science at University of Arizona.