Identification of critical factors for assessing the quality of restaurants using data mining approaches
| Date | 09 December 2019 |
| Published date | 09 December 2019 |
| Pages | 952-969 |
| DOI | https://doi.org/10.1108/EL-12-2018-0241 |
| Author | Ahsan Mahmood,Hikmat Ullah Khan |
| Subject Matter | Information & knowledge management |
Identification of critical factors
for assessing the quality
of restaurants using data
mining approaches
Ahsan Mahmood
Department of Computer Science, COMSATS Institute of Information Technology,
Attock, Pakistan, and
Hikmat Ullah Khan
Department of Computer Science, COMSATS University Islamabad –Wah Campus,
Wah Cantt, Pakistan
Abstract
Purpose –The purpose of this paper is to apply state-of-the-artmachine learning techniques for assessing
the quality of the restaurants using restaurantinspection data. The machine learning techniques are applied
to solve the real-world problems in all sphere of life. Health and food departments pay regular visits to
restaurants for inspection and mark the condition of the restaurant on the basis of the inspection. These
inspectionsconsider many factors that determine the condition of the restaurantsand make it possible for the
authoritiesto classify the restaurants.
Design/methodology/approach –In this paper, standardmachine learning techniques, support vector
machines, naïve Bayesand random forest classifiersare applied to classify the critical level of the restaurants
on the basis of features identified duringthe inspection. The importance of different factors of inspectionis
determined by using featureselection through the help of the minimum-redundancy-maximum-relevanceand
linear vectorquantization feature importance methods.
Findings –The experiments are accomplishedon the real-world New York City restaurant inspection data
set that contains diverse inspection features. The results show that the nonlinear support vector machine
achieves better accuracythan other techniques. Moreover, this researchstudy investigates the importance of
different factors of restaurant inspection and findsthat inspection score and grade are significant features.
The performance of the classifiers is measured by using the standardperformance evaluation measures of
accuracy,sensitivity and specificity.
Originality/value –This research uses a real-worlddata set of restaurant inspection that has, to the best
of the authors’knowledge,never been used previously by researchers. The findingsare helpful in identifying
the best restaurants and help finding the factorsthat are considered important in restaurant inspection. The
resultsare alsoimportant in identifying possible biases in restaurant inspectionsby the authorities.
Keywords Classification, Restaurants, Quality assurance, Feature selection, Machine learning
Paper type Research paper
1. Introduction
The research domain of machine learning is applied so that computers can learn from the
given data and can solve real-world problems. These machine learning techniques use
available data to learn patterns of data and solve tasks. The machine learning tasks are
usually divided into several broad categories, including supervised, semi-supervised,
unsupervised and reinforcement learning. Among the tasks solved by these machine
EL
37,6
952
Received10 December 2018
Revised25 April 2019
6 August2019
26August 2019
Accepted5 September 2019
TheElectronic Library
Vol.37 No. 6, 2019
pp. 952-969
© Emerald Publishing Limited
0264-0473
DOI 10.1108/EL-12-2018-0241
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0264-0473.htm
learning techniques, classificationis one of the most important. Classification is the process
of classifying the availabledata into two or more groups on the basis of their characteristics.
This process is highlyuseful in many domains and has applications in a number of fields.
Restaurants are inspected for quality-control purposes by health and food departments
on a regular basis (Jones et al., 2004). Factorsthat are involved in determining the quality of
restaurants, generally, are determined by four dimensions, namely, price fairness, physical
environment, food and service quality (Ryu and Lee, 2017). Restaurant inspections are
considered an important source of information to evaluate a restaurant’s food and quality
(Choi et al.,2013). During these inspections, restaurants are tested for several quality
measures. In the modern era, many state-of-the-art technologies have been introduced for
restaurant inspections that helpto produce a better assessment of restaurants (Jin and Lee,
2014). After restaurants are inspected for different quality-control measures, decisions are
made by the authorities to mark the restaurantsas either “not critical”or “critical”–in other
words, whether the restaurantis left open for business or is closed for public safety reasons.
These inspections contain a number of factors that help in determining the condition of
restaurants. However, because of the lack of systematic approaches for decision-making, it
sometimes becomes difficult for the authorities to classify the restaurant condition on the
basis of the inspection. There are a number of applications of restaurant classification that
can help all the stakeholders of restaurants. These can also help the restaurant owner to
maintain their ratings, the inspection teams to automatically classify restaurants based on
data and visitors to choose the best restauranton the basis of its quality. The exploration of
the factors that define the quality of the restaurantsusing machine learning algorithms still
remains a research problem, and this research work aims to rank such factors and classify
the restaurantsbased on these attributes using widely used machinelearning.
In this paper, the domain of restaurants is studied and a solution to classify restaurant
conditions on the basis of their inspectionis proposed. The researchers approach the domain
of determining restaurant quality through the help of user reviews and, to the best of our
knowledge, there is no work solelyfocused on classifying restaurant conditions on the basis
of their inspections. The lack of work in this domain is highly influenced by the
unavailability of the inspection data set that just recentlybecame available. The inspection
records consist of a high number of factors; therefore, the classification is approached by
first selecting the best features among the other factors. Thefeatures are selected by using
the minimum-redundancy-maximum-relevance (mRMR) feature selection and learning
vector quantization (LVQ) feature importancemethods. Afterward, the binary classification
of restaurant conditions is approached through the help of linear and nonlinear support
vector machines (SVMs) and the naïve Bayes classifier. The experiments reveal the
important factors in restaurant inspections. The experimental results are evaluated by the
means of standard performance measurement techniques in terms of accuracy, specificity
and sensitivity. Although acceptable classification accuracy is achieved, the final results
show that the decisive variable of restaurant inspection does not depend upon many
factors and cannot efficientlybe predicted because of the inconsistency in the decisions.The
results also reveal that score and grade are the most important factors in restaurant
inspections havingthe highest impact upon the decision relativeto the other factors.
The rest of the paper is organized as follows. Section 2 discusses the proposed research
methodology and the basic concepts of different models used in this paper. Section 3
discusses all the details about the proposed experimental setup and the techniques used to
measure the performance of the experiments. Section 4 shows the performed experiments,
the results and the findings of the research.Section 5 reports conclusions and future work.
Assessing the
quality of
restaurants
953
To continue reading
Request your trial