You are on page 1of 3

Abstract:

The research aims to control the power of machine learning topredict food safety inspection
scores for restaurants. The general question is whether machine learning models can
proficiently anticipate inspection scores by using examining various attributes connected to a
food establishment's adherence to safety regulations.

The research methodology covers the collection and processing the data derived from food
safety inspections. The dataset covers a wide choice of information such as, serial numbers,
facility names, violation codes, and more. Machine learning algorithms like Gradient Boosting,
Decision Trees, K-Nearest Neighbors, Random Forest, Naive Bayes, and Support Vector
Machines were used to create the predictive models. It was divided into training and testing
stages, which enables careful assessment of model performance through metrics like accuracy
and mean squared error.

The result of the study can help improve the safeguarding of public health and enforcing its
regulatory compliance. The use of machine learning models during the inspection can show
efficiency which will in turn permit targeted interventions in the establishment with lower
predicted scores.

Keywords:
1. Machine Learning- Computer-based learning from data to make predictions.
2. Food Safety - Ensuring safe food practices and compliance.
3. Predictive Models- Mathematical tools for anticipating outcomes.
4. Inspection Scores - Ratings based on safety inspections.
5. Regulatory Compliance - Adherence to safety rules and guidelines.

Introduction:

Food safety is an important concern for the consumers and the authorities. In previousyears,
incidents that is related to food safety has risen, creating concerns about the quality of the foods
in different establishment. Analyzing the data that we provided, can give important information
into the food safety sector. By using machine learning algorithms, the research aims to create a
predictive model that can determine the potential of food safety in an establishment.

Topic Overview:

The dataset consists of inforamtion regarding food establishments, their serial numbers, activity
dates, facility names, and attributes related to food safety ciolation and inspection scores. It has
a total of 1,418 rows and 14 columns. It ranges from object to integer, with two numerical
columns: 'points' and 'score.'.
Data was split into training and testing in which several models were used:Gradient Boosting,
Decision Tree, K Nearest Neighbors, Random Forest, Naive Bayes, and Support Vector
Machine. Random Forest got 94.84% accuracy which is the highest, while Support Vector
Machine is the lowes with 18.54%.

Topic Discussion:

The dataset shows an important resource for analyzing and improvement of food safety inside a
food establishment. It shows a varied information like, serial numbers, activity dates, facility
names, violation codes, and more. Upon analyzing the data, it is clear that it has underwent
prepocessing with categorical variables, making it more malleable for machine learning
techniques (C.Aldrich.2013)

The main objective when using this kind of data is to predict the score variable. To do this, the
dataset needs to be divided into two sets, training and testing. This is a common style in
machine learning, which enables the user to assess the algorithms perform according to the
scores of the food establishments. Machine learning encompasses various techniques
employed to analyze data, make informed choices, and forecast outcomes.

The study used a variety of classifiers like, Gradient Boosting, Decision Tree, K Nearest
Neighbors, Random Forest, Naive Bayes, and Support Vector Machine. Random Forest has
risen to the top with an accuracy rate of 94.84% this shows the efficacy of this method in
handling complex data such as food safety inspections.

But the classifier chosen is not just an activity in data analysis rather it has an important
implication. Random forest classifier getting such high rate accuract can be a special tool for
health departments and food safety to make their inspection short but efficient, diminishing the
risk of illness from food. Also a confusion matrix was used in order to further evaluate the
performace of the classifiers. This will tell us how the model determines the scores and it will
help gove insights into the prediction errors like, false positives and false negatives.

Conclusion:

This study highlights that the model Random Forest can be used to determine high-risk food
establishments and guide food safety more effciently, which will in turn improve public health
and safety. It also invites deeper researh into the practical application of same models to help
improve food safety inspection and processes. So in conclusion, the creation and successful
application of machine learning based model for food safety shows a positive step forward in
ensuring the well-being of the customer.
Reference:

Uçar, M. K., Bozkurt, M. R., Bilgin, C., & Polat, K. (2017). Automatic detection of respiratory
arrests in OSA patients using PPG and machine learning techniques. Neural Computing and
Applications, 28(10).

https://www.kaggle.com/

You might also like