p-ISSN: 2301-5373

e-ISSN: 2654-5101

Jurnal Elektronik Ilmu Komputer Udayana

Volume 10 No. 1, August 2021

Optimization of Bali Tourism Recommendations Based on Personal Motivation of Tourists Using the Naive Bayes Algorithm

I Gusti Ngurah Agung Widiaksa Putraa1, I Gusti Agung Gede Arya Kadyanana2, Ida Bagus Gede Dwidasmaraa3

Informatic Engineering, Udayana University Campus UNUD Bukit Jimbaran, Indonesia 1Widiaksaputra.WIP@gmail.com 2gungde@unud.ac.id 3dwidasmara@unud.ac.id

Abstract

In the recovery of the tourism sector in Bali due to COVID-19, a solution is needed with the aim of making tourists more interested in having a vacation in Bali. One of the solutions that can be offered is optimizing the tourist recommendations on the island of Bali, because so far tourists only get travel recommendations from travel agents and guides who usually recommend favorite tourist destinations, and sometimes guides recommend tourist attractions according to their personal wishes or goals. making tourists less optimal in enjoying tourist attractions in Bali. Optimization of Bali Tourism Recommendations Based on Tourist Personal Motivation Using the Naive Bayes Algorithm, is one solution to optimize tourism recommendations in Bali, where tourist recommendations are taken based on tourist characteristics using the Naïve Bayes Algorithm. In this study the authors used 180 training data, and the results of this study indicate that the personal motivation of tourists who are processed using the Naïve Bayes algorithm is feasible to use for tourism recommendations in Bali.

Keywords: Recommendation Optimization, Personal Motivation, Naïve Bayes.

  • 1.    Introduction

During the COVID-19 pandemic, many industries have suffered losses, especially the tourism sector, according to the Bali Tourism Industry Association (GIPI) Data, the potential losses of the tourism sector in Bali from leisure and mice reach USD 9 billion or around IDR 140 trillion exchange rate of IDR 15,639). According to the provincial government of Bali, the spread of the COVID-19 issue has resulted in a decrease in the number of tourists to Bali, especially Chinese tourists [1]. Thus, a solution is needed to restore the tourism sector in Bali, considering that Bali is one of the best tourist destinations in Indonesia. However, improving the economy in Indonesia as a whole certainly takes time, especially considering that the COVID-19 pandemic has yet to be confirmed.

Based on these problems, given the possibility of economic recovery due to the COVID-19 pandemic, a solution that can be supported by the government is urgently needed, with the aim of making tourists more interested in having a vacation in Bali. One of the solutions that can be offered

is optimizing the Tourism Recommendations on the island of Bali, because so far tourists only get travel recommendations from guides or travel agents, where the travel recommendations are only based on favorite tourist attractions, sometimes even guides recommend tourist attractions according to their wishes, or personal goals, even though with the choice of unfounded tourist recommendations, it is likely that tourists are less than optimal in enjoying tourist attractions in Bali, this can be proven in the Market Analysis book and the 2019 Archipelago Tourist Satisfaction Index, the impression of tourist attractiveness by domestic tourists visiting Bali (35.3%) is very good, (62.6%) is good, (1.3%) is quite good, (0.8%) is very poor, while the level of satisfaction of domestic tourists visiting Bali (27.5 %) very satisfied, (68.4%) satisfied, (3.4%) quite satisfied, (0.6%) less satisfied, (0.1%) very dissatisfied [2], thus it can be concluded that a solution is needed. i to optimize the impression and satisfaction of tourists on tourist attractions in Bali.

Therefore, the authors propose an idea, namely Optimization of Bali Tourism Recommendations Based on Personal Motivation of Tourists Using the Naive Bayes Algorithm. With the concept of optimized Bali tourism recommendations, which is how to optimize it based on the personal motivation of tourists such as tourist characteristics to provide a set of recommendations to related tourists, as well as how to process tourist characteristics data using the Naïve Bayes algorithm to produce tourist recommendations. The personal motivation of tourists that I use comes from the attributes of age, origin, gender, interest. The data that will be used comes from the 2019 National Tourist Characteristics Survey in Bali Province. With these attributes it can certainly produce tourism recommendations that are in accordance with the character of tourists, so that tourists travel more optimally in Bali, besides that it is hoped that tourist visits in Bali will be more evenly distributed.

  • 2.    Reseach Methods

Optimization of Bali Tourism Recommendations based on Tourist Personal Motivation was developed using several designs, methods and algorithms. The following is a discussion of the Bali Tourism Recommendation Optimization method, algorithm. The Naive Bayes algorithm is used to classify tourist data, with the aim of providing the greatest possibility, the suitability of tourist attractions based on the characteristics of the related tourists, whose simple concept is to compare tourist characteristics data related to tourist characteristics data that have previously determined tourist attractions, so recommendations are given from the classification results.

Data

Training k___________

  • 2.1.    Optimization

Optimization is a process to achieve ideal results or optimization (the effective value that can be achieved). Optimization can be interpreted as a form of optimizing something that already exists, or designing and making something optimally [3].

  • 2.2.    Tourism

The definition of tourism is a trip that is carried out individually or in groups to a certain place which is carried out repeatedly or going around, either in a planned or unplanned manner which can produce a total experience for the perpetrator [4].

There are several factors that influence tourists in traveling to a place, according to the journal [5].

There are 24 factors that are taken into consideration by tourists visiting a tourist spot, namely: personal motivation, availability of attractive tourist objects and products, advice and recommendations from travel agents, information obtained about tourist destinations from tourism and travel organizations agents, recommendations from friends, friends or relatives (words of mouth), political conditions, security and technology of tourism destinations, environmental hygienic conditions of tourist destinations, special promotions from tourism organizations, climate and weather of tourist destinations, attitudes, opinions and perceptions of tourist destinations, lifestyle of these tourism consumers, current knowledge of tourist destinations, hobbies and interests as well as past experiences of tourism consumers who have visited a particular tourist destination, commitment to

family, income levels, consumer personalities, transportation and road access, and supporting infrastructure for accommodation facilities i, hotel food and beverage and others. Middleton, Fyall & Morgan (2009) provide a model for the decision-making process for tourism consumers, there are 3 models, namely, Process, Stimulation and Response.

RESPON

Tourist Behavior What ?, Where ?, How Much and Often?

Figure 2, Tourism decision model, STIMULI, PROCESSING, RESPONSE

In Figure 2 above, you can see a model that influences tourists' decisions in traveling to a place, starting from STIMULI which contains marketing communications, and other sources of information about the destination tourist attractions. Then proceed to PROCESSING which involves the characteristics of tourists both culturally, socially, personally, and psychologically in receiving and posting information, so that a tourist decision-making process occurs. So that tourists can provide RESPONSE regarding the tourist information received, in the form of questions about tourist attractions that tourists want to know more about.

Based on the journal, it is concluded that 15 main factors from the 24 factors tested, in the influence of decision making on tourists in choosing tourist attractions, namely:

Faktor

rJidak

Ya

Ranking

Motivasi Persona)

83

317

1

Sikap. Opini & Persepsi

86

314

2

Ketersediaan Obyek & Produk Wisala

87

313

3

Kepribadian

87

313

3

Transportasi & ALses Jalan

91

309

5

Kondisi Kebersihan & Lingkungan

93

307

6

Words of Mouth

94

306

7

Pengetahuan ttg Destinasi Wisata

96

304

8

Fasilitas Utama & Pendukung

98

302

9

Tingkat Pendapatan

99

301

10

Kesehatan Personal

1∞

300

Il

Event & Hiburan

IOI

299

12

Cuaca & Iklim

103

297

13

Hobi & Ketertarikan

103

297

13

Kondisi Politik & Keamanan

112

288

15

Sumber: Data diolah, Februari 2015

Figure 3, Main Factors

  • 2.3.    Personal Motivation

Personal motivation, according to Swarbrooke & Horner (2007) consists of physical (relaxation, health, comfort, etc.), social (visiting friends and family, meeting work partners, doing things that bring prestige, etc.), the desire to know the culture, customs, traditions and other regional arts, selfactualization, and security. From the explanation regarding tourist factors in making decisions about tourist attractions you want to visit above. The author uses the factors of age, gender, origin, and interest as attributes in the classification process to provide recommendations for tourists to choose tourist attractions. The influence of the factors of age, gender, and origin is the environment that tourists have so that it affects the personal motivation of tourists in making decisions about choosing tourist attractions. In addition, there is an attribute of interest which is one of the main factors for tourists in choosing tourist attractions, namely those related to hobbies and interests.

  • 2.4.    Algoritma Naive Bayes

The Naïve Bayes Classifier is a classification method rooted in the Bayes theorem. The classification method uses probability and statistical methods proposed by the British scientist Thomas Bayes, which predicts future opportunities based on previous experience, so it is known as Bayes' Theorem. The main characteristic of this Naïve Bayes Classifier is a very strong assumption (naive) of the independence of each condition / event. The Naive Bayes Classifier performs very well in comparison to other classifier models. This is evidenced in the journal "Naïve Bayes vs. Decision Trees vs. Neural Networks in the Classification of Training Web Pages "says that" Naïve Bayes Classifier has a better level of accuracy than other classifier models "[6]. The advantage of using this method is that this method only requires a small amount of training data to determine the parameter

estimates required in the classification process. Since it is assumed to be an independent variable, only the variance of a variable in a class is needed to determine the classification.

  • 3.    Result and Discussion

The following is the data that the authors use as training data in this study, where this training data comes from the results of a survey of the characteristics of domestic tourists in the province of Bali in 2019, which has been carried out by the Bali Provincial Tourism Office, in collaboration with the University's Culture and Tourism Research Center. Udayana.

H ħ∙ =

File Home Insert Page Layout Formulas Data Review

f1

A

A

I C

D

E

1

Kelamin □ Usia □ Asal     □ Minat    □ Destinasi Wisata B

2

Perempuan

16-25

Yogyakarta

Berbelanja

Pantai Kuta

3

Laki-Laki

26-55

Jawa Timur

Melihat-lihat

Pantai Pandawa

4

Laki-Laki

16-25

Jakarta

Berbelanja

Tanah Lot

5

Perempuan

16-25

Jawa Barat

Berbelanja

GWK

6

Perempuan

16-25

Yogyakarta

Berbelanja

Toya Bungkah

7

Perempuan

16-25

Jawa Barat

Berbelanja

Pantai Kuta

S

Laki-Laki

26-55

Jawa Timur

Berbelanja

Pantai Kuta

9

Perempuan

16-25

Yogyakarta

Berbelanja

Pantai Pandawa

10

Laki-Laki

26-55

Jawa Tengah Berbelanja

Pantai Pandawa

11

Perempuan

26-55

Jawa Tlmur

Berbelanja

Pantai Pandawa

12

Laki-Laki

16-25

JawaTengah Berbelanja

Pantai Kuta

13

Perempuan

16-25

JawaTengah Berbelanja

Pantai Kuta

14

Perempuan

16-25

JawaTengah Berbelanja

Pantai Kuta

15

Laki-Laki

16-25

JawaTengah Berbelanja

Pantai Kuta

16

Laki-Laki

26-55

Jawa Timur

Berbelanja

Pantai Kuta

17

Laki-Laki

26-55

Sumatera

Berbelanja

Pantai Kuta

18

Laki-Laki

26-55

Sumatera

Berbelanja

Pantai Kuta

19

Perempuan

26-55

Sumatera

Berbelanja

Pantai Kuta

20

Perempuan

16-25

Jakarta

Petualangan

Pantai Kuta

21

Perempuan

26-55

Sumatera

Berbelanja

Pantai Kuta

22

Perempuan

16-25

Jawa Timur

Petualangan

Pantai Kuta

23

Laki-Laki

16-25

Sumatera

Berbelanja

Pantai Kuta

24

Laki-Laki

26-55

Jawa Timur

Melihat-lihat

Pantai Pandawa

25

Perempuan

16-25

Sumatera

Berbelanja

Pantai Kuta

26

Perempuan

26-55

Sumatera

Berbelanja

Pantai Kuta

27

Laki-Laki

16-25

Sumatera

Berbelanja

Pantai Kuta

28

Laki-Laki

16-25

Sumatera

Berbelanja

Pantai Kuta

29

Laki-Laki

16-25

Jawa Timur

Berbelanja

Pantai Kuta

E

: p

HT

M wβ

φ 9

  • Figure 4,    Training Data

In this case, currently the author only uses 180 training data, to apply the Naïve Bayes algorithm in the Optimization of Tourism Recommendations, there are 4 attributes & 1 label that the author uses, each value in the attribute consists of :

-Sex (Male, Female)

-Age (<= 15,16-25,26-55)

-Asal (Jakarta, West Java, Central Java, East Java, Sumatra, Yogyakarta)

  • - Interests (Shopping, Sightseeing, Adventure)

Whereas the label contains the values:

-Tourism Destinations (GWK, Kuta Beach, Pandawa Beach, Tanah Lot, Toya Bungkah)

Which is where the author only uses data from domestic tourists, and from 1190 authors use 180

random data.

Here is the calculation of Naïve Bayes:

, , , P(Vlw) «™ - -Jm


For example:

Table 1, Example Cases

Kelamin

Usia

Asal

Minat

Destinasi Wisata

Laki-Laki

16-25

Jawa Tengah

Melihat-lihat

????????

Table 2, Examples of the Naïve Bayes Classification Process

Perhitungan Naive Bayes

Hasil

Label

16/38 * 16/38 * 4/38 * 19/38 * 38/180

0.00196984

GWK

32/68 * 41/68 * 10/68 * 35/68 * 68/180

0.0081134

Pantai Kuta

10/19 * 7/19 * 4/19 * 11/19 * 19/180

0.00249469

Pandawa

20/45 * 31/45 * 0/45 * 24/45 * 45/180

0

Tanah Lot

2/10 * 6/10 * 0/10 * 5/10 * 10/180

0

Toya Bungkah

To eliminate the 0 in naïve Bayes calculations we need the Laplacian Correction algorithm, along with

the calculations:                        m√÷l

Pi τι-∖-k

Table 3, Example of a Laplacian Correction Process

Perhitungan Naive Bayes + Laplacian Correction

Hasil

Label

17/76 * 17/76 * 5/76 * 20/76 * 38/180

0.000182875

GWK

33/136 * 42/136 * 11/136 * 36/136 * 68/180

0.000606093

Pantai Kuta

11/38 * 8/38 * 5/38 * 12/38 * 19/180

0.000267289

Pandawa

21/90 * 32/90 * 1/90 * 25/90 * 45/180

0.000064014

Tanah Lot

3/20 * 7/20 * 1/20 * 6/20 * 10/180

0.00004375

Toya Bungkah

Based on the calculations from table 2 to table 3, a tourism recommendation can be given, namely

Kuta Beach. The results of the analysis using the python language program are:

JLipytfir PariwisataNaiveBayeS Last Checkpoint 4hours ago JautosaveC)                                                           ^® Logout

File Edit View Insert Cell Kernel Widgets Help                                                                                     Trusted ≠k ∣ Python 3 C

when a Id array was expected. Please change the shape of y to (n-samplesj ), for example using ravel(). return f(**kwarg≡)

In [628]: from sklearn import metrics

In [629]: print("Accuracy: '∖metrics.accuracy_score(y testj y_pred))

Accuracy: 0.3611111111111111

In [632]: predicted = gnb.predict([[Θ,0λ2j1]]) hasil ≡ print("predicted value: "j predicted) predicted value: [1]

In [631]: df.replace(hasil)

Out[631]:

Kelamin  Usia        Asal       Minat DestinasiWisata  Kelamin encode Usia encode Asal encode  Minat encode Destinasi Wisata encode

O  Perempuan  16-25   Yogyakarta   Berbelanja       Pantai Kuta                 1             0 5              01

1   Laki-Laki 26-55  Jawa Timur Melihat-lihat  Pantai Pandawa              0           13            12

2 I aki-l aki 16-75 Jakarta Rarbelania          Tanah^r^                 0              0              0                03

  • Figure 5,    Results of Program Analysis

In python the author uses the GaussianNB () library to calculate the Naïve Bayes algorithm. In this program, each value and label is simplified into numeric variables, then it can be read :

Table 4, Translation of Program Results

Kelamin

Usia

Asal

Minat

Destinasi Wisata

Laki-Laki

16-25

Jawa Tengah

Melihat-lihat

????????

0

0

2

1

????????

In accordance with Figure 5 the result is: variable 1 which means Kuta Beach Based on the python program analysis in Figure 5, it is concluded that the resulting accuracy is between 80% of training data and 20% of testing data, with an accuracy of 0.3611111111, or it can be concluded that 36%.

  • 4.    Conclusion

Based on this research, it can be concluded that, the Naïve Bayes Algorithm can be used to maximize travel recommendations based on the Personal Motivation of Tourists, and based on the results of this study, the Application of Travel Recommendations using the Naïve Bayes Algorithm based on the Personal Motivation of Tourists has the potential to be done, and in the future it will be very beneficial for the tourism industry.

References

[1]     Pemerintah     Provinsi     Bali,     “www.baliprov.go.id”,     13    September

https://www.baliprov.go.id/web/pers-release/#.

2020.

[2]   Dinas   Pariwisata   Provinsi   Bali,   “disparda.baliprov.go.id”,   13   September

https://disparda.baliprov.go.id/buku-analisa-pasar-wisatawan-2019/.

2020.

[3]      KBBI      Daring,      https://kbbi.kemdikbud.go.id/,      14      September

https://kbbi.kemdikbud.go.id/entri/optimasi.

2020.

[4]      KBBI      Daring,      https://kbbi.kemdikbud.go.id/,      14      September

https://kbbi.kemdikbud.go.id/entri/pariwisata.

2020.

  • [5]    Yusendra, “ANALISIS FAKTOR-FAKTOR YANG MEMPENGARUHI KEPUTUSAN PEMILIHAN DESTINASI WISATA BAGI WISATAWAN DOMESTIK NUSANTARA”, Vol. 01, No.1, Januari 2015.

  • [6]    Xhemali, J. Hinde, G. Stone, “Naïve Bayes vs. Decision Trees vs. Neural Networks in the Classification of Training Web Pages”, 4 (1), pp. 16-23, 2009.

This page is intentionally left blank

90