Model Geographically Weighted Univariate Weibull Regression Studi Kasus: Indikator Pencemaran Sungai di Surabaya

Santoso, Fitriarma Putri (2015) Model Geographically Weighted Univariate Weibull Regression Studi Kasus: Indikator Pencemaran Sungai di Surabaya. Undergraduate thesis, Institut Technology Sepuluh Nopember.

Text 1313201051-Master Thesis.pdf - Published Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Metode yang digunakan dalam memodelkan hubungan antara variabel
respon dan variabel prediktor adalah analisis regresi. Suatu model regresi dimana
variabel respon (Y) berdistribusi weibul dengan variabel prediktor X , X ,..., X k 1 2
adalah model regresi weibul. Seperti hal nya analisis regresi linier, hubungan
antara variabel prediktor dan variabel respon pada analisis regresi weibul
dianggap konstan untuk setiap lokasi geografis, sehingga penaksir parameter yang
didapat juga bersifat global untuk setiap lokasi. Geographically Weighted
Univariate Weibull Regression (GWUWR) merupakan bentuk lokal dari regresi
weibul dan merupakan metode statistik yang digunakan untuk menganalisis data
spasial. Univariat berarti jumlah variabel respon adalah satu. Penelitian ini
bertujuan untuk menaksir parameter model GWUWR, mendapatkan statistik uji
GWUWR dan menerapkannya pada kasus riil. Data yang digunakan merupakan
data konsentrasi Chemical Oxygen Demand (COD) sungai di Surabaya dan faktor
yang mempengaruhinya. Dalam menaksir parameter menggunakan metode
Maximum Likelihood Estimation (MLE) dengan dibantu metode iterasi Newton
Raphson, sedangkan untuk mendapatkan statistik uji menggunakan Maximum
Likelihood Ratio Test (MLRT). Faktor lokasi tidak berpengaruh terhadap
pemodelan COD sehingga faktor-faktor yang berpengaruh di setiap lokasi hampir
sama. Faktor-faktor yang berpengaruh terhadap Chemical Oxygen Demand
(COD) di sungai Surabaya berdasarkan model GWUWR antara lain lebar sungai
(X1), kedalaman sungai (X2), kecepatan air (X3) dan debit sungai (X4).
=========================================================================================================
The method used to modelling the relationship between the response
variable and the predictor variable is regression analysis. Response variable (Y),
weibull distribution, with the predictors variable X , X ,..., X k 1 2 is weibull
regression model. Such as linear regression analysis, the relationship between
predictor variables and response variable in weibull regression assumed to be
constant for each geographical location, then the parameter estimation is also
obtained global for each location. GWUWR is a statistical method for analyzing
spatial data and a local form of weibull regression. Univariate means the number
of the response variable is only one. This study aims to estimate the parameters of
GWUWR model, getting a test statistic of GWUWR and applying it to the real
case. The data that used is data of concentration of Chemical Oxygen Demand
(COD) in Surabaya river and their the factors that influence. Estimate parameters
are use Maximum Likelihood Estimation (MLE) with helped by Newton Raphson
iteration method, whereas to get the test statistic using Maximum Likelihood
Ratio Test (MLRT). Location factors have not given influence to COD modelling,
hence every factors influenced in every location are almost the same. Factors
influence in Chemical Oxygen Demand (COD) in Surabaya river based on
GWUWR models are the wide of the river (X1), the depth of the river (X2), the
speed of the water (X3) and river discharge (X4).

Item Type:	Thesis (Undergraduate)
Additional Information:	RTSt 519.536 San m
Uncontrolled Keywords:	Distribusi weibul, Geographically Weighted Univariate Weibull Regression (GWUWR), Chemical Oxygen Demand (COD)
Subjects:	Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression
Divisions:	Faculty of Mathematics and Science > Statistics > 49101-(S2) Master Thesis
Depositing User:	Mr. Tondo Indra Nyata
Date Deposited:	18 Nov 2019 08:39
Last Modified:	18 Nov 2019 08:39
URI:	http://repository.its.ac.id/id/eprint/71862

Actions (login required)

View Item