NutriPal-Recipe-Recommendation-System-

Project Overview

In the rapidly growing health and wellness industry, individuals are increasingly seeking practical solutions to make informed dietary choices and improve their overall well-being. However, navigating the vast array of diet plans, meal delivery services, and health apps can be overwhelming. Stakeholders in this industry face the critical challenge of providing personalized and accurate nutrition recommendations that meet individuals' unique needs and preferences.

A significant problem is the lack of tailored nutrition guidance available in the market. Existing solutions often offer generic diet plans that do not consider individual factors such as age, gender, body composition, dietary restrictions, and cultural preferences. Consequently, individuals may experience frustration and disappointment when these solutions fail to deliver the desired results, leading to a decline in motivation and a higher likelihood of abandoning their healthy eating goals.

Furthermore, the fast-paced nature of modern lifestyles presents another obstacle. Many individuals struggle to find the time and energy required to research, plan, and prepare nutritious meals regularly. This often results in resorting to unhealthy eating habits, negatively impacting their overall health and well-being.

By providing accurate and personalized nutrition recommendations, stakeholders in the health and wellness industry can differentiate their offerings, enhance customer satisfaction, and foster long-term adherence to healthy eating habits. Additionally, utilizing advanced technologies and user-friendly interfaces can create a competitive advantage and position stakeholders as leaders in the market.

Problem Statement

The global prevalence of obesity and other diet-related chronic diseases is increasing. This is due in part to the increasing availability of unhealthy foods and the difficulty people have in making healthy food choices.

A food recommendations system could help people make healthier food choices by providing personalized recommendations based on their individual needs and preferences. This could help people improve their diet and reduce their risk of developing chronic diseases.

Solution Statement

Lack of personalized nutrition recommendations: Existing solutions often rely on generic dietary guidelines that are based on population-level data. This approach can be effective for some individuals, but it may not be optimal for everyone. For example, generic guidelines may not take into account an individual's genetic makeup, which can play a role in determining how their body responds to different foods. Additionally, generic guidelines may not be tailored to an individual's lifestyle or health goals. For example, someone who is trying to lose weight may need different recommendations than someone who is trying to manage a chronic disease.

Time and effort constraints for meal planning: Busy lifestyles can make it difficult for individuals to find the time and energy to plan and prepare healthy meals. This can be especially challenging for people who work long hours or have young children. Additionally, meal planning can be time-consuming and complex, especially if an individual has food allergies or intolerances.

Objectives

Main objective.

Develop a Food/Recipe Recommendation System that suggests nutritious food to individuals and promotes a healthy lifestyle.

Specific Objectives.

Identify the key features and factors that impact an individual's overall health and determine which ones should be incorporated into the food recommendation system.
Clean and preprocess the nutrition data available in the dataset and combine it with external data sources to create a comprehensive nutrition database that can be used by the recommendation system.
Develop and implement recommendation algorithms that can generate personalized food recommendations based on the user's individual characteristics, such as age, gender, degree of physical activity, locally available foods, and dietary customs.
Integrate the recommendation algorithms and chatbot into a user-friendly and intuitive interface that allows users to easily access and interact with the system.
Deploy the food recommendation system and conduct user testing to gather feedback and identify areas for improvement.

Metrics Of Success

Our recommender system will be considered successful if it meets the following criteria:
Have a recall score of 80% and a lower MSE Value.
Have coverage of around 90%. This means that the model is able to recommend a wide variety of nutritious foods and recipes to users

Data Understanding

This project will include 3 datasets

Foreign Recipes
Nutrition
Kenyan Local Food Recipes

The recipe data set was obtained from here . It contains a list of 231636 rows of recipes and 12 columns.

name - Recipe name
id - Recipe ID
minutes - Minutes to prepare the recipe
contributor_id - User ID who submitted this recipe
submitted - Date recipe was submitted
tags - Food.com tags for recipe
Nutrition - Nutrition information (calories (#), total fat (PDV), sugar (PDV) , sodium (PDV) , protein (PDV) , saturated fat (PDV) , and carbohydrates (PDV))
n_steps - Number of steps in the recipe
steps - Text for recipe steps, in order
description - User-provided description
ingredients - List of ingredient names
n_ingredients - Number of ingredients

The nutrition dataset was obtained from here.

This dataset contains information on approximately 8.8 thousand types of food. The dataset includes various features related to the nutrition value of each food item per 100gram serving. There are 75 features in total, you can find features like calories, vitamin_d, zink, protein, lactose. As you can see feature names are very self-explanatory, so a description is not provided.

The Kenyan Dataset was obtained from here The dataset contains information about recipes, including their names, cooking time (in minutes), ingredients, steps, serving size, and nutritional information. There are 218 entries (recipes) in the dataset.

The ' Serving' column represents the serving size of each recipe. The ' calories', ' total fat (PDV)', ' sugar (PDV)', ' sodium (PDV)', ' protein (PDV)', ' saturated fat (PDV)', and ' carbohydrates (PDV)' columns provide nutritional information for each recipe, expressed as percentages of the re *recommended daily value (PDV).

MODELING

We used collaborative filtering and Content-based filtering to provide recommendations based on a user's preference. The models utilized datasets containing nutrition information and calories for various food items. The recommender allows users to select a specific food item with calorie levels and provide a list of recommendations that meet the requirements.

Collaborative filtering analyzes user behavior and preferences to recommend items that other similar users have liked. This approach is based on the assumption that users who have similar tastes and preferences in the past will have similar preferences in the future. Collaborative filtering can be helpful in recommending food items based on the choices of users with similar dietary preferences and calorie requirements.

Content-based filtering, on the other hand, focuses on the attributes or characteristics of items themselves. Our final model was the content-based filtering approach that involved analyzing the nutrition information and calorie levels of various food items to recommend similar items that meet the user's specific requirements.

EVALUATION

The evaluation of the recommender system provided valuable insights into its performance. Content-Based Filtering Model:

Precision: 0.2
Recall: 0.5
Hit Rate: 0.2

The content-based filtering model achieved a precision of 0.2, indicating that 20% of the recommended recipes were actually liked by the user. The recall value of 0.5 suggests that all the liked recipes were successfully recommended. The hit rate of 0.2 implies that only 20% of the recommended recipes were liked by the user.

Collaborative Filtering Model:

Precision: 0
Recall: 0
Hit Rate: 0

The collaborative filtering model did not yield any meaningful results in this case, resulting in all evaluation metrics being 0.

DEPLOYMENT

We used Streamlite to deploy our model. The model takes in User preferences and Calorie levels to provide recommendations for recipes as shown below.

CONCLUSION

In conclusion, the recipe recommender system successfully utilized content-based filtering to provide personalized recipe recommendations. By analyzing the nutrition information and calorie levels of food items, the system was able to suggest similar recipes that matched the user's preferences and dietary requirements. The evaluation results highlighted the effectiveness of the content-based filtering approach in terms of precision, recall, and we will use it in our deployment.

RECOMMENDATIONS

To enhance the user experience and promote healthy eating habits, we suggest incorporating the following features into the recommendation system:

User Profile: Allow users to create profiles and input their specific dietary preferences, restrictions, and goals. This information can be used to further personalize the recipe recommendations.
Meal Planning: Integrate a meal planning feature that suggests a balanced set of recipes for each day or week, considering the user's calorie requirements and dietary preferences. This feature can help users plan their meals in advance and maintain a healthy eating routine.
Recipe Filtering: Provide options to filter recipes based on specific dietary needs, such as vegetarian, vegan, gluten-free, or low-carb. This allows users to find recipes that align with their specific dietary preferences and restrictions.
Nutritional Information: Include detailed nutritional information for each recipe, including calories, macronutrients (protein, fat, carbohydrates), and key vitamins or minerals. This information can help users make informed decisions about the nutritional content of the recipes they choose.
User Feedback and Ratings: Implement a user feedback system where users can rate and provide feedback on the recipes they try. This feedback can be used to improve the recommendation system and provide more accurate and relevant suggestions in the future.

mugangasia / nutripal-recipe-recommendation-system- Goto Github PK