Statistical analysis (Exploratory Data Analysis) of the nutritional values (macronutrients) of recipes of different types of diets in order to generate a statistical description of the diets.

Alexis Aguilar [Student of Bachelor's Degree in "Tecnologías para la Información en Ciencias" at Universidad Nacional Autónoma de México UNAM]: alexis.uaguilaru@gmail.com

Project developed for the subject "Descriptive and Inferential Statistics" for the class taught in semester 2025-2.

Project under MIT License

The purpose of this work is to present the process carried out to perform the statistical analysis of the nutritional values (macronutrients) provided by recipes of the following diets: DASH (Dietary Approaches to Stop Hypertension), keto, mediterranean, paleo and vegan. The main focus is to answer whether there is a significant nutritional difference between the different diets, or equivalently whether each diet follows its own trend or preferences of the foods that are consumed and prepared.

Making use of different techniques and tools of descriptive and inferential statistics to perform a statistical analysis of the macronutrients consumed in the different recipes with the objective of describing their macronutrient consumption patterns and trends according to the diet to which they belong, and testing whether there is a significant difference between the different patterns and trends reported in each diet based on the nutritional contributions (macronutrients).

The dataset is taken from [1]. The dataset contains information on the nutritional values of carbohydrates, proteins and fats of recipes from different international cuisines and diets.

In Exploratory Data Analysis the different ideas and comments related to the statistical analysis of the recipes based on the context of the diet to which they belong are presented. This generates a better interpretation of the observed statistics and how they are related based on the products and foods representative of each diet, which leads to a more consolidated knowledge about the trends and patterns that follow the diets presented in [2], and which are statistically reinforced.

This last result contributes significantly to the General Aim of the project, that is, evidence is being generated to support the hypothesis that each diet distinguishes itself from others on the basis of nutritional contributions. This can be best observed by looking at the distribution of macronutrients in each diet.

In Hypothesis Testing the hypothesis test for the difference between diets based on nutritional contributions or macronutrient distributions is presented. The Kolmogorov-Smirnov test is used to detect differences in macronutrient distributions between two diets, from the results obtained it is concluded that there is a significant difference between all diets, and to measure this difference it is proposed to use the mean of the values of the test statistics for each macronutrient; the following means are obtained for each pair of diets:

As can be seen in the table, there are diets that are more similar than others, this is related to the products and foods consumed, and to the trends that according to what is exposed in Exploratory Data Analysis, so it becomes that, from a certain point of view, they behave in a similar way. Specifically, it could be figured that the diets are grouped in the following way and explanation:

• DASH and Vegan: Both diets tend to be rich in the consumption of foods of vegetable origin, so this implies diets rich in carbohydrates and fats; this means that they are also deficient in proteins. And these two patterns can be seen in their distributions.

• Paleo and Mediterranean: Although they consume different foods, their nutritional contributions are similar, they tend to be more balanced or diversified among the different macronutrients. The latter means that their differences are not notable.

• Keto: Unlike other diets, this diet has a high consumption of fats so that proteins and carbohydrates, mainly, have a deficient consumption. This fact characterizes this diet, which is a very different trend from other diets.

First it has to clone the repository and move to the project directory:

git clone https://github.com/alexisuaguilaru/Analysis_Nutritional_Information
cd Analysis_Nutritional_Information

Preferably using a virtual environment of Python, install the libraries required for the project with:

pip install -r requirements.txt

With this, one can run and view the Marimo notebooks with the following command:

marimo run StatisticalAnalysis/ExploratoryDataAnalysis.py
marimo run StatisticalAnalysis/HypothesisTesting.py

• Python
  • Marimo
  • Pandas
  • Seaborn
  • SciPy
  • NumPy

• [1] Diets, Recipes And Their Nutrients. Kaggle. https://www.kaggle.com/datasets/thedevastator/healthy-diet-recipes-a-comprehensive-dataset
• [2] F. F. Marvasti, "Popular Diets and Health", Culinary Medicine