Pemodelan identifikasi penyebaran kebakaran hutan dan lahan berdasarkan kondisi cuaca dengan mlr, rf, dan svm studi kasus: pulau Sumatera = Modeling identification of the spread of forest and land fire based on weather conditions with mlr, rf, and svm case study: Sumatera island

Callista, Michelle (2022) Pemodelan identifikasi penyebaran kebakaran hutan dan lahan berdasarkan kondisi cuaca dengan mlr, rf, dan svm studi kasus: pulau Sumatera = Modeling identification of the spread of forest and land fire based on weather conditions with mlr, rf, and svm case study: Sumatera island. Bachelor thesis, Universitas Pelita Harapan.

[img]
Preview
Text (Title)
Title.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (90kB) | Preview
[img]
Preview
Text (Abstract)
Abstract.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (188kB) | Preview
[img]
Preview
Text (ToC)
ToC.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (478kB) | Preview
[img] Text (Chapter1)
Chapter1.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (745kB)
[img] Text (Chapter2)
Chapter2.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (1MB)
[img] Text (Chapter3)
Chapter3.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (1MB)
[img] Text (Chapter4)
Chapter4.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (4MB)
[img] Text (Chapter5)
Chapter5.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (290kB)
[img]
Preview
Text (Bibliography)
Bibliography.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (533kB) | Preview
[img] Text (Appendices)
Appendices.pdf
Restricted to Repository staff only
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (1MB)

Abstract

Kebakaran hutan dan lahan merupakan salah satu bencana yang biasa terjadi di Indonesia. Hal ini tidak dapat dihindari mengingat panjangnya musim kemarau serta faktor cuaca lainnya seperti suhu panas yang cukup tinggi. Dampak dari kebakaran hutan dan lahan mempengaruhi banyak aspek mulai dari rusaknya alam, gangguan aktivitas yang disebabkan oleh asap, hingga kerugian ekonomi. Salah satu cara yang dapat dilakukan untuk mengurangi kebakaran hutan dan lahan secara efektif adalah dengan melakukan upaya pencegahan. Bentuk upaya pencegahan meliputi pengembangan sebuah model komputasi penyebaran kebakaran untuk memprediksi lokasi kebakaran berdasarkan kondisi cuaca. Tahap pengerjaan penelitian terdiri atas lima tahap yang terdiri atas data understanding, data preparation, modeling, evaluation, dan deployment. Pengembangan model diawali dengan mengumpulkan data kejadian dan lokasi kebakaran serta kondisi cuaca seperti arah angin dan temperatur. Data-data tersebut kemudian melalui proses analisis dan persiapan sebelum digunakan dalam pembangunan model. Model yang digunakan meliputi Multinomial Logistic Regression, Random Forest, dan Support Vector Machine. Ketiga model telah banyak diterapkan dalam melakukan pemodelan kejadian kebakaran dan menunjukkan kemampuan prediksi yang kuat dalam probabilitas kejadian kebakaran. Pembangunan model dilakukan dengan menggunakan kondisi cuaca untuk memprediksi koordinat garis lintang dan bujur dari lokasi kebakaran. Perbandingan model kemudian dilakukan untuk mencari model yang paling cocok dengan data kebakaran yang dimiliki. Hasil dari penelitian menunjukkan penggunaan model random forest memberikan hasil error yang paling rendah dibandingkan dengan model lainnya dengan nilai mean squared error 6,54 derajat, root mean squared error 2,55 derajat, dan mean absolute error 1,77 derajat. Berdasarkan hasil tersebut dapat disimpulkan bahwa model random forest dapat dengan baik melakukan prediksi menggunakan data kebakaran hutan dan lahan serta data kondisi cuaca. Visualisasi data kebakaran hutan dan lahan, data cuaca, dan hasil prediksi tiap model telah diluncurkan dalam bentuk aplikasi web menggunakan Streamlit untuk diakses secara publik. / Wildfires is one of the common disasters that happens in Indonesia. This event in unavoidable given the length of the dry season and other weather factors such as high temperatures. The impact of wildfires affects many aspects, starting from the destruction of nature, disruption of activities caused by smoke, to economic losses. One way that can be done to reduce wildifres effectively is to take preventive measures. Forms of prevention efforts include the development of a computational model of fire spread to predict the location of fires based on weather conditions. The research work stage consists of five stages consisting of data understanding, data preparation, modeling, evaluation, and deployment. Model development begins with collecting data on the occurrence and location of fires as well as weather conditions such as wind direction and temperature. These data then go through the process of analysis and preparation before being used in model development. The models used include Multinomial Logistic Regression, Random Forest, and Support Vector Machine. These models have been widely applied in modeling fire events and shows a strong predictive ability in the probability of fire occurence. Model development is done by using weather conditions to predict the latitude and longitude coordinates of the fire location. Comparison of these models is then carried out to find the model that best fits the existing wildfire data. The result of the study show that the use of random forest model gives the lowest error compared to other models with a mean squared error of 6.54 degrees, a root mean squared error of 2.55 degrees, and a mean absolute error of 1.77 degrees. Based on these results, it can be concluded that the random forest model is suitable for making predictions using data on wildfires as well as data on weather conditions. Visualization of wildfire data, weather data, and the prediction results of each model have been deployed in the form of a web application using Streamlit to be accessed publicly.

Item Type: Thesis (Bachelor)
Creators:
CreatorsNIMEmail
Callista, MichelleNIM01082190007michellecalista94@gmail.com
Contributors:
ContributionContributorsNIDN/NIDKEmail
Thesis advisorMurwantara, I MadeNIDN0302057305made.murwantara@uph.edu
Thesis advisorLazarusli, IreneNIDN0317097501irene.lazarusli@uph.edu
Uncontrolled Keywords: wildfires, weather condition, multinomial logistic regression, random forest, support vector machine, Sumatera
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: University Subject > Current > Faculty/School - UPH Karawaci > School of Information Science and Technology > Informatics
Current > Faculty/School - UPH Karawaci > School of Information Science and Technology > Informatics
Depositing User: Michelle Callista
Date Deposited: 10 Feb 2023 05:12
Last Modified: 10 Feb 2023 05:12
URI: http://repository.uph.edu/id/eprint/54158

Actions (login required)

View Item View Item