Personalized Food Recommendation System by Using Machine Learning Models
Personalized Food Recommendation System by Using Machine Learning Models
Personalized Food Recommendation System by Using Machine Learning Models
ISSN No:-2456-2165
Abstract:- A recommender system’s primary function is particular diseases they may be experiencing. This is done
tomake recommendations to users. So, in this study, we by looking at the foods that contain one of the main
developed a “Food Recommendation System” that components and how healthy they are rated [3].The
suggests foods based on the age provided and a rating recommender system is the foundation of this endeavor.
greater than 4. This work connects users to recipe ideas, Tools that use supervised learning for recommendations.
ingredient lists, and cooking times, serving as a hub for When using supervised learning, we already know the
kitchen-related information.This study has the dataset user’s needs and use our ML model to make
with 12 features, namely name of the dish,style of recommendations [4].Nutrition is getting the food and
cooking,category, calorie count, flavor profile, cooking nutrients needed for development and good health. We
skills, course, healthy, preparation time, ingredient become weaker, get sicker, and so on.In the worst-case
count, rating and age, where data cleansing is done and scenario, we might die if we do not consume [5]. We
is divided into 80/20 training and test data, respectively. cannotregularly subject our bodies to taxing mental and
Decision trees, random forest, K-Nearest Neighbour, physical activities and fall short of developmental
and logistic regression models were applied to the benchmarks.An ideal diet must be balanced for nutrition
training dataset, and the expected value for the test and well-being. It protects against chronic,
dataset was then produced. For each of these models, a noncommunicable diseases like diabetes, heart disease, and
combined confusion matrix and categorization report cancer [6]. All living things, including humans, plants,
were created.Based on the confusion_matrix and animals, and other creatures, depend on them to survive.
classification_report numbers, validate the results with Nutrition recommendation systems are one of the most
Precision, Recall, F1-score, and Accuracy for each widely used technological innovations in nutrition
model.Calculated the misclassification rate for each knowledge. (NRS). They are being researched as a
algorithm and showed the most accurate model.This potential tool to assist users in forming better eating
dataset consists of 488 records. routines and achieving their goals [7] . A recommender
system can anticipate users’ preferences for unrated goods
Keywords:- Decision Tree,K-Nearest Neighbour,Logistic and recommend new products.
Regression, and Random Forest.
A. Activity Diagram
I. INTRODUCTION The progression of one action through a system or
process is shown in an“activity diagram,” a flowchart in the
They all consider what to eat, where to eat, how much Unified Modeling Language (UML). It is used to describe
dietary worth a food has, and whether it will help them lose the different dynamic components of a system and is
weight or build muscle, among other things. [1] Thus, this known as a “behavior diagram” because it specifies what
method enables the user to make decisions in these should happen in the modeled system [8].
complex information spaces quickly. A meal
recommendation is a personalized service that works In this study, the initial action for a user is to log in
wonders for helping with diet planning. This recommender and input an age range, as shown in Figure 1.The food
may abuse the meal’s nutritional value to justify its items are sorted, after which suggestions are generated.
suggestions. The application aims to provide users with a
platform to learn about their favorite foods’ nutritional
values and find new foods to try. Anyone who wants to
lose weight is worried about their health, or is dealing with
a particular health issue can gain something from this.
Systems for calorie counting are getting much attention as
they have begun to replace the old with and stylus
techniques.These programs offer services and educational
resources that influence users to change their behavior [1].
Due to the widespread use of these calorie-counting offices,
these systems have access to a vast amount of client
preference data that can be used to tailor intuitive features
and boost user engagement with the system and the slim-
down plan [2]. The user of this food recommendation app
can find their favored foods and learn more about their
nutritional value in light of any health conditions or
A. Numpy
A popular Python package for numerical processing is
Fig. 1: Activity Diagram called NumPy. It works exceptionally well with arrays and
matrices. It offers multidimensional arrays and different
II. LITERATURE REVIEW mathematical functions that can be used on them in a high-
performance implementation. Due to their contiguous
The study articles taught how recommender systems memory allocation and ability to vectorize operations,
filter out the necessary information andoffer a suggestion to NumPy arrays are more effective than Python lists, making
a user based solely on their current needs. In this them ideal for scientific computing, data analysis, and
personalized and non-personalized recommender system, machine learning apps.
they are essentially types. In personalized recommender
systems, the information is first gathered from the user, for Numerous linear algebra operations are available in
instance, by conducting surveys or distributing a NumPy, including matrix multiplication, decomposition,
questionnaire to understand the user’s likes and dislikes and eigenvalue computations. A fast Fourier transform
and recommendations [9]. In contrast, in non-personalized (FFT) tool for signal processing and spectral analysis is
recommender systems, the information is given based on also included in NumPy. The work is extensively used in
observation of likes. Also, the top trending hits at the time, various industries, including data science, engineering,
for example, Netflix displays the top 10 shows in each finance, and scientific computing.
region, and one more type is based on the one thing that is
watched. Others are decided using machine learning Several built-in functions in NumPy, such as
algorithms, categorized, and displayed as similarities [10]. reshaping, slicing, and indexing, make manipulating and
changing array data simple. For additional analysis and
According to Anna Gatzioura and Miquel Sanchez, visualization, they can also be coupled with other data
the recommendation system aims to provide users with structures, such as pandas DataFrames. NumPy is a vital
valuable and pertinent material while actively using the instrument for Python’s effective and efficient numerical
site. (items).[11,12] Recommendation systems have computing.
become increasingly popular in recent years. In the middle
of the 1990s, the first publication on collaborative filtering B. Pandas
appeared after recommendation systems emerged as a A popular Python tool for data analysis and
popular research subject. To filter and retrieve info, the manipulation is called Pandas. It offers flexible, high-
recommendation system technology is used [13]. These performance data structures like Series (1-dimensional) and
technologies also help e-commerce websites and other DataFrame (2-dimensional) for managing tabular data.
networks increase their sales. These systems are a piece of NumPy, which Pandas is based upon, enables the practical
work that helps customers find the products they prefer and computation of sizable datasets.
connects them to the services and products they are
interested in [14]. Recommendation systems are a type of Filtering, combining, grouping, and pivoting are just a
machine learning technology that falls under unsupervised few of the functions that Pandas offers for cleaning,
machine learning models, in which the data is not transforming, and analyzing data. The library also provides
categorized [15]. solutions for managing duplicate data and handling missing