Ambary, Umar El - Farouk (2026) Analisis Faktor Yang Mempengaruhi Jumlah Kecelakaan Lalu Lintas Di Provinsi Indonesia Menggunakan Metode Regresi Data Count. Other thesis, Institut Teknologi Sepuluh Nopember.
|
Text
5002211142-Undegraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy |
Abstract
Kecelakaan lalu lintas merupakan salah satu permasalahan serius di Indonesia yang terus menunjukkan tren peningkatan setiap tahunnya dengan penyebab melibatkan berbagai faktor. Penyebab kecelakaan lalu lintas melibatkan berbagai faktor yang kompleks, mulai dari aspek manusia, kendaraan, kondisi jalan, hingga faktor lingkungan dan ekonomi, sehingga diperlukan analisis komprehensif untuk mengidentifikasi faktor-faktor yang berpengaruh signifikan terhadap jumlah kecelakaan. Dalam konteks tersebut, regresi data count digunakan sebagai metode regresi statistik yang sesuai untuk memodelkan variabel respon berupa bilangan bulat non-negatif yang merepresentasikan jumlah kejadian suatu peristiwa. Penelitian ini memodelkan jumlah kecelakaan lalu lintas di 34 provinsi Indonesia pada tahun 2023 dengan menggunakan metode regresi data count, yaitu Regresi Poisson, Generalized Poisson Regression dan Negative Binomial Regression. Regresi Poisson bersifat equidispersi, yaitu mengasumsikan bahwa nilai rata-rata sama dengan varians data. Apabila asumsi equidispersi tidak terpenuhi dan model Regresi Poisson menunjukkan adanya overdispersi maupun underdispersi, digunakan model alternatif berupa Generalized Poisson Regression (GPR) yang mampu menangani masalah overdispersi dan underdispersi atau Negative Binomial Regression (NBR) yang mampu menangani masalah overdispersi. Data variabel yang dianalisis bersumber dari Badan Pusat Statistik (BPS) dan Kementerian PUPR. Estimasi parameter model regresi data count menggunakan metode Maximum Likelihood Estimation dan Newton-Raphson, yang kemudian dilanjutkan dengan pengujian parameter secara serentak dan parsial. Diakhiri dengan uji devians serta evaluasi model menggunakan Akaike Information Criterion (AIC). Berdasarkan hasil penelitian, model Negative Binomial Regression yang paling optimal dengan memiliki nilai rasio devians terhadap derajat bebas (df) yang paling mendekati satu, yaitu sebesar 1,3917, dan nilai AIC yang terkecil, yaitu sebesar 763,8718, sehingga dapat mengatasi permasalahan ini. Model Negative Binomial Regression untuk jumlah kecelakaan lalu lintas di setiap provinsi Indonesia tahun 2023 adalah µˆi = exp(12,3841 + (-6,320)x4T ) yang menyatakan bahwa jumlah kecelakaan lalu lintas di Indonesia tahun 2023 pada provinsi ke-i adalah exp(12,3841) seiring dengan exp(-6,320) berkurangnya presentase kondisi jalan rusak.
=====================================================================================================================================
Traffic accidents are one of the serious problems in Indonesia that continue to show an increasing trend each year, with causes involving various factors. The causes of traffic accidents involve various complex factors, ranging from human factors, vehicles, road conditions, to environmental and economic factors, so a comprehensive analysis is needed to identify factors that significantly influence the number of accidents. In this context, count data regression is used as an appropriate statistical regression method to model the response variable in the form of a non-negative integer representing the number of occurrences of an event. This study models the number of traffic accidents in 34 provinces of Indonesia in 2023 using count data regression methods, namely Poisson Regression, Generalized Poisson Regression, and Negative Binomial Regression. Poisson Regression assumes equidispersion, meaning that the mean is equal to the variance of the data. When the equidispersion assumption is not satisfied and the Poisson Regression model indicates the presence of overdispersion or underdispersion, alternative models are employed, namely Generalized Poisson Regression (GPR), which is capable of handling both overdispersion and underdispersion, or Negative Binomial Regression (NBR), which is capable of handling overdispersion. The analyzed variables are obtained from the Central Bureau of Statistics (BPS) and the Ministry of Public Works and Public Housing (PUPR). Parameter estimation of the data count regression models is carried out using the Maximum Likelihood Estimation method and the Newton–Raphson algorithm, followed by simultaneous and partial parameter testing. The analysis is concluded with a deviance test and model evaluation using the Akaike Information Criterion (AIC). Based on the research results, the Negative Binomial Regression model is the most optimal, as it has a deviance-to-degrees-of-freedom ratio (df) closest to one, namely 1,3917, and the smallest AIC value, namely 763,8718, thereby addressing this problem effectively. The Negative Binomial Regression model for the number of traffic accidents in each province of Indonesia in 2023 is given by µˆi = exp(12,3841 + (-6,320)x4T ), which indicates that the number of traffic accidents in Indonesia in 2023 in the i-th province is exp(12,3841), along with exp(-6,320) for a decrease in the percentage of damaged road conditions.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Regresi Poisson, Generalized Poisson Regression, Negative Binomial Regression, Kecelakaan Lalu Lintas, Equidispersi, Poisson Regression, Generalized Poisson Regression, Negative Binomial Regression, Traffic Accidents, Equidispersion. |
| Subjects: | Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression Q Science > QA Mathematics > QA401 Mathematical models. |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis |
| Depositing User: | Umar El - Farouk Ambary |
| Date Deposited: | 27 Jan 2026 07:08 |
| Last Modified: | 27 Jan 2026 07:08 |
| URI: | http://repository.its.ac.id/id/eprint/130496 |
Actions (login required)
![]() |
View Item |
