Vol. 4, No. 2, Juli, 2023 : Estimasi

One of the goals of national development is to reduce poverty. Poverty is included in the phenomenon of spatial heterogeneity because it can be shown by the varying economic conditions in each region. The statistical modeling method developed for data analysis takes into account regional factors namely Geographical Weighted Logistic Regression (GWLR). The parameter estimator of the GWLR semiparametric model used in this study was obtained using the Maximum Likelihood Estimation method. In GWLR, the assumption that must be fulfilled is the absence of multicollinearity. One method for dealing with multicollinearity is ridge regression involving the addition of a constant bias . The results obtained were the MSE value of the parameter estimator with the ridge method (707.77) smaller than the GWLR model before using the ridge (715.88). This shows that the ridge method is more effective if there are multicollinearity problems.

Logistic regression ordinal is a regression model that can explain the relationship between predictor variables in the form of categorical data or continuous data with response variable is more than two categories with a scale of measurement that is level or sequence. In ordinal logistic regression, the frequency of occurrence in each response category is often very different, so it will affect the model's accuracy. Therefore, this study will model ordinal logistic regression with a dispersion of location effects, then applied to the nutritional status data of toddler in 2019 at the Pekkae Puskesmas, Barru Regency. The results obtained show that the ordinal logistic regression model with the dispersion of location effects is better than the usual ordinal logistic regression model for predicting the nutritional status data for toddlers in 2019 at Pekkae Puskesmas, Barru Regency based on deviance values. The factors that influence the nutritional status of toddler based on TB/U are gender, age, and height.

Gross Regional Domestic Product (GRDP) often shows spatial patterns. In a spatial perspective, spatial effects consist of of spatial dependence and spatial heterogeneity. To address the problems, this study uses spatial autoregressive quantile regression/SARQR model. SARQR is a method that combines Spatial Autoregressive (SAR) modeling with quantile regression. There are two methods that can be used to estimate the parameters of the SARQR model, namely Two Stage Quantile Regression (2SQR) and Instrumental Variable Quantile Regression (IVQR). The simulation results showed that IVQR method is better than 2SQR method. IVQR provides a smaller value and variance of bias. Furthermore, IVQR method is applied to Java’s GRDP data on 2019. The results showed that the number of workers significantly influences Java’s GRDP. The highest quantile verification skill score (QVSS) value is 0.713 when τ =0.75. It means that in the 75% quantile modeling, the model can describe the GRDP diversity of 71.3%.

Climate change that occurs is often used to predict future climate conditions. For future climate predictions it is only possible to use climate models. One of the climate models used to predict climate conditions is the Global Circulation Models (GCM). GCM represents global climatic conditions but not on a regional or local scale. The approach that has been widely used to bridge the difference in scale is statistical downscaling. Large-scale GCM data allows for multicollinearity. estimation liu regression and principal component regression is used to solve the multicollinearity problem. In addition, dummy variables based on k-means and average linkage are used in the model to overcome the heterogeneous variance of residue. There are 4 dummy variables in the cluster technique. In this paper, Liu k-means regression model parameter estimation method is the best model.

If the variable is a discrete random variable with Poisson distribution, the data analysis must fulfill the equidispersion assumption. In reality, these assumptions are not fulfilled because the variance is greater than the mean which is called overdispersion. Overdispersion in data can occur due to the proportion of excess zero values in these variables. To estimate the parameters, the MLE method can be used on data that has a certain distribution by maximizing the likelihood function, it obtained is implicit or nonlinear so that it cant be solved analytically. To get the numerical solution, it solved by using the EM algorithm. The estimation results of the ZIGP distribution parameters are used to create control chart limits for the 2016 Neonatal Mortality Rate data in Makassar with limits of , , and . The  chart ARL value is , which is greater than the chart ARL value, which is  which indicates that the  chart is better at detecting outliers.

Earthquake is a natural phenomenon that is random in nature because its occurrence depends on time so that earthquakes are seen as a Nonhomogeneous Poisson Process. In this study, the Nonhomogeneous Poisson process was applied to estimate the number of earthquakes on the island of Sulawesi. The data used in this study is the occurrence of earthquakes on Sulawesi Island from January 2018 to December 2020 sourced from the Meteorology, Climatology and Geophysics Agency (BMKG) Region IV Makassar. The results of this study indicate that earthquakes that occur from one month to the next do not affect each other other than that the value of the intensity of the earthquake in each interval (month) is not the same, so that the estimated incidence of earthquakes on the island of Sulawesi with a strength of more than 5.0 SR is obtained. on 1 to 8 July 2021 is about 14 earthquakes with a standard deviation of about 3 times and the probability of an earthquake is 0.10537.

Indonesia in 2022, will experience a shift in adaptation to recovery from the pandemic as well as rising global commodity prices due to the impact of the Ukraine-Russia war. The government in its efforts to deal with this situation, one of which is by transforming data into one data through the 2022 Social Economic Registration (Regsosek) as a requirement for social protection system reform. However, in practice, Research and Research has become quite a public concern, where the content is almost the same as previous surveys conducted by BPS, which raises questions about the effectiveness of this survey. This study aims to determine the sentiments of each opinion on social media Twitter regarding 2022 Social Security. This research implements the K-Nearest Neighbor (K-NN) method to analyze sentiment in tweets. Data obtained from Twitter by scrapping. The polarity percentage results from the tweets obtained are dominated by negative opinions. The best application of the K-Nearest Neighbor (K-NN) algorithm is using the parameter k = 3. The model built shows very good performance with an accuracy of 96%, a recall of 100%, and a precision of 0,96%.

Value at Risk (VaR) is statistical method used in risk analysis in stock investments. Stock returns that are not normally distributed cause the risk calculation to be less precise, so to overcome this, the copula method can be used. Copula is a method based on dependencies between variables. The most commonly known copula family is the Archimedean copula which consists of the Clayton, Frank, and Gumbel copula. VaR is expected to be a feasible method to use, so it is important to perform backtesting. In this research, we use data on the daily closing price of PT. Adaro Energy Tbk and PT. Bukit Asam Tbk May 11, 2020 until June 15, 2022. The best copula based on the smallest Empirical copula value is Frank copula. VaR estimates for the 90%, 95%, and 99% confidence levels respectively were 2.688%, 3.545%, and 5.014%. The higher the confidence level, the VaR value is also higher. Based on backtesting results, VaR with Frank copula method is valid at 90%, 95%, and 99% confidence levels.

Maternal mortality is the death of a woman who is pregnant, giving birth and childbirth to the pregnancy or its handler. Maternal mortality in East Java Province still quite high with the highest number of deaths in 2021 is Jember Regency. The purpose of this paper is to determine risk factors that cause death in an effort to reduce the number of maternal deaths. Method used for the analysis of risk factors for maternal mortality is survival analysis with the Cox Proportional Hazard model. Survival analysis purpose to assess the relationship of predictor variables to survival time to determine maternal survival. Cox Proportional Hazard model is one of the models in survival analysis that is often used. Selection of the best model for Cox Proportional Hazard is carried out to determine the factors that have a significant effect. The best model is done by selecting the smallest AIC value backwards. Parameter significance test on the best model was carried out simultaneously and partially. Results obtained for maternal mortality factors in Jember Regency are anemia status and parity.

Gross Regional Domestic Product (GRDP) is an important indicator of economic growth in a region. The success of regional economic growth is said to be good if the GRDP in an area has a significant effect on that area. However, economic growth in East Nusa Tenggara (NTT) has not been optimal. This is caused by economic inequality in NTT which differs between districts/cities. Therefore, the aim of this research is to find out what factors influence GRDP in NTT using Geographically Weighted Panel Regression (GWPR). The data used is sourced from the Central Statistics Agency (BPS) website. The results of the study describe that GRDP in NTT is divided into 12 groups with the adaptive bisquare kernel function and the coefficient of determination is 83.73%. The independent factors that influence GRDP in NTT are the Construction Expensive Index (IKK) and Area Area (LW) in the construction sector, the Human Development Index (IPM) in the Social sector, the Number of Poor population in the population sector, and the Literacy Rate (AMH) in the education sector. Meanwhile, the health sector did not affect GRDP in NTT.

The aim of the study was to determine the pattern of relationship between the sales turnover variables of cake vendors, the use of wheat flour, and gender. Knowing the best equation model that shows the relationship pattern of cake vendor sales turnover variables, the use of wheat flour, and gender. Determine the cells that cause the relationship between variables to be dependent. The data analysis method used in this research is the Linear Log Model with the best model using the Backward method. The data in this research is secondary data from a survey at PT. BOGASARI in November 2022. The conclusions obtained in this study are that there is a relationship between the variable sales turnover of traders every day and the use of wheat flour for golden carriages. The best model showing the pattern of relationship between the sales turnover variables of cake vendors, the use of carriage golden flour, and gender is log . Male traders whose daily turnover is < Rp. 500,000.00 tend to use Kereta Kencana flour, while female traders whose turnover is < Rp. 500,000.00 per day tend not to use Kereta Kencana flour.

The rapid development of technology causes the use of mobile phones and the need for pulses to increase. East Lombok is the area with the largest population in NTB and high users of information technology. East Lombok also has an internet network or smooth communication signal, which shows that there are many providers in the area. To see the types of providers that are widely used in Masbagik District, East Lombok, taking into account the largest population, a classification is made of whether these providers are in demand or not using the Naïve Bayes method. This study aims to determine the classification results and the accuracy of the sales classification of credit providers. The data is split into two categories: training data (90%) and testing data (10%). According to the findings of the study, 225 of the 309 testing data were correctly classified. The resulting APER value is 27.2%, which indicates that the accuracy of the classification results using the Naïve Bayes method is 72.8%. An AUC value of 0.804 was also obtained, which means that the accuracy of the classification of selling pulse providers that are in demand, moderately in demand, and not in demand was sufficient.