AIC Value Research Articles

Geographically Weighted Regression with Kernel Weighted Function on Poverty Cases in West Java Province

Spatial regression analysis is a form of regression model that considers spatial effects. Geographically weighted regression (GWR) is the spatial regression methods that can be used to deal with the problem of spatial diversity. This method generates local model parameter estimates for each observation location. The application of spatial statistics can be done in all areas such as the problem of poverty. Poverty can be influenced by factors of proximity between regions, so that in determining the poverty factor, the proximity factor of the region cannot be ignored. West Java Province is a province with the largest population, so this study aims to model the poverty data in West Java Province by incorporating spatial effects. The weighting function used for the GWR model is the function of the fixed and adaptive kernels. The analysis results show that the fixed exponential kernel function has the smallest cross validation (CV) value, so the weighting matrix used in the model is determined by the exponential kernel function. The largest value and the smallest AIC value are owned by the GWR model with an exponential kernel function. Based on the results obtained by the the ANOVA table to test GWR's global goodness, the GWR model is more effective than global regression. Therefore, the GWR model is the best model when it used in West Java’s poverty cases. The effect of each explanatory variable on the percentage of poverty varies in each district/city in West Java Province.

Forecasting outbreak of COVID-19 in Turkey; Comparison of Box–Jenkins, Brown’s exponential smoothing and long short-term memory models

The new coronavirus disease (COVID-19), which first appeared in China in December 2019, has pervaded throughout the world. Because the epidemic started later in Turkey than other European countries, it has the least number of deaths according to the current data. Outbreak management in COVID-19 is of great importance for public safety and public health. For this reason, prediction models can decide the precautionary warning to control the spread of the disease. Therefore, this study aims to develop a forecasting model, considering statistical data for Turkey. Box-Jenkins Methods (ARIMA), Brown’s Exponential Smoothing model and RNN-LSTM are employed. ARIMA was selected with the lowest AIC values (12.0342, -2.51411, 12.0253, 3.67729, -4.24405, and 3.66077) as the best fit for the number of total case, the growth rate of total cases, the number of new cases, the number of total death, the growth rate of total deaths and the number of new deaths, respectively. The forecast values of the number of each indicator are stable over time. In the near future, it will not show an increasing trend in the number of cases for Turkey. In addition, the pandemic will become a steady state and an increase in mortality rates will not be expected between 17–31 May. ARIMA models can be used in fresh outbreak situations to ensure health and safety. It is vital to make quick and accurate decisions on the precautions for epidemic preparedness and management, so corrective and preventive actions can be updated considering obtained values.

ARMA-GARCH model for value-at-risk (VaR) prediction on stocks of PT. Astra Agro Lestari.Tbk

PT. Astra Agro Lestari Tbk (AALI) is one of the plantation companies with the largest market capitalization in Indonesia. AALI stocks traded on the stock exchange have fairly fluctuating value and volatility of stock returns are not constant (heteroskedastic). One of the risk measurements that can be used to predict the risk of stock investing is Value-at-Risk (VaR). In conditions that are heteroskedastic stock returns, risk prediction can be done with the VaR ARCH/GARCH and VaR ARCH/GARCH combination model. Empirical studies were carried out on AALI stocks for the period of August 2, 2012 until October 1, 2019. The results obtained showed that the best model was ARIMA (0,0,1)-GARCH (1,2) with AIC value of -4.9793 and MSE of 0.00005. At the 95% trust level, the VaR ARCH/ARCH value was -0.3464.

Analysis of (co) variance components and estimation of breeding value of growth and production traits in Dahlem Red chicken using pedigree relationship in an animal model.

Variance and covariance components of growth and production traits were analyzed employing REML animal model to assess the Dahlem Red (PD-3) chicken population for direct additive genetic, maternal effects and to estimate the estimated breeding value (EBV), genetic parameters, genetic trends and rate of inbreeding (ΔF) utilizing seven generation’s data. The generation and hatch had significant (P≤0.01) effect on the body weight at 0 day (BW0), 2 (BW2), 4 (BW4) and 6 weeks (BW6) and shank length at six weeks of age (SL6). The average least squares means (LSM) for BW6 and SL6 were 273.93±0.62 g and 53.97±0.05 mm, respectively. All the production traits were significantly (P≤0.01) influenced by generation and hatch. The average LSM for age at sexual maturity (ASM), egg production up to 40 weeks (EP40) and egg mass up to 40 weeks (EM40) were 168.82±0.25 d, 72.60±0.41 eggs and 4.21±0.07 kg, respectively. Model 5 with additive direct, maternal genetic, maternal permanent environmental and residual variance components was the best for BW0, BW2 and BW4 based on the AIC values obtained in WOMBAT. Model 4 was the best model for BW6, SL6, ASM, EP40 and EM40 with additive direct, maternal permanent environmental and residual variance components. Maternal effects were higher during early age, decreased with age, and remained present until 20 weeks of age. The heritability (h2) estimates were low to moderate in magnitude for all the growth traits and ranged from 0.02±0.03 to 0.19±0.03. The maternal heritability was high at hatch (0.35±0.06), decreased gradually until 4th week (0.02±0.01) and ceased afterwards. The heritabilities of EP40 (0.11±0.03) and EM40 (0.12±0.04) were low. The direct additive genetic correlations (ra) between BW2, BW4, BW6 and SL6 were high and positive (P≤ 0.05). The additive genetic and maternal permanent environmental correlation between EP40 and EM40 were high and positive (P≤ 0.05). The EBV of EM40 was significant (P≤ 0.05) with 0.48 kg/generation in PD-3 chicken at the end of the seventh generation. The EBV of EP40 showed an increasing trend with a genetic gain of 1.87 eggs per generation. The average inbreeding coefficient of the population was 0.019 and average ΔF was 0.007 over the last seven generations of selection. The EBV trends for primary and associated traits showed linear trends in the desired direction and negligible inbreeding.

Factors associated with consecutive and non-consecutive crashes on freeways: A two-level logistic modeling approach

A consecutive crash consists of a primary crash and one or more secondary crashes that occur subsequently in a short period of time within a certain distance. It often affects a relatively large area of road space and the traffic disruption created can be difficult for traffic managers to control and resolve. This study identifies the factors delineating a primary crash that results in secondary crashes within a minute from a regular crash that does not result in any secondary crashes. Random-effects, random-parameter and two-level binary logistic regression models are applied to data collected on 8779 crashes on the freeway network of the Guizhou Province, China in 2018, of which 299 are consecutive crashes. According to the AIC values, the two-level logistic model outperforms the other two models. Rear-end primary crashes have a significant random effect varying across road segments on the occurrence of consecutive crashes. Various crash types (rear-end, roll-over and side-swipe), tunnel crash and foggy weather are positively associated with the possibility to cause subsequent consecutive crashes, whereas single-vehicle crash, truck involvement and the time periods with poorer natural lighting are less likely to incur consecutive crashes. Recommendations are provided to minimize the possibility of the occurrence of consecutive crashes on a freeway.

A Bayesian approach for Generalized Linear Model Using Non-local Prior (Case Study: Poverty Status in East Java)

Generalized Linear Model (GLM) is an extension of the linear regression model. GLM is useful for determining the effect of predictor variables to the response variable that is a member of the exponential distribution family. This study uses a binary response variable having a Binomial distribution. This modelling uses the logit link function. This study aims to evaluate the performance of the Bayesian GLM modelling using local prior distribution and non-local prior distribution. The non-local prior density in this study was pi-MOM. The results of the research on modelling poverty status in East Java using the local prior distribution produced a predictor variable that had a significant effect on poverty status in East Java. Meanwhile, modelling using non-local prior distribution produced three predictor variables that significantly affected the status of poverty in East Java. The goodness criteria of the model showed that the Bayesian GLM modelling using the prior non-local distribution was the best model with the smallest misclassification and the lowest AIC value.

An Alternative Procedure to Produce a P-Spline Small Area Estimation Model Based on Partial Residual Plot and Significance Test of Spline Term

This study provides an alternative procedure to produce the penalized spline small area estimation (P-Spline SAE) model by plotting each of auxiliary variable and variable interest as a simple nonlinearity identification and performing the iterative procedure: by estimating the model, producing partial residual plots to check model adequacy and also identify the nonlinearity, and testing the significance of spline term using Restricted Likelihood Ratio Test (RLRT). These procedures are applied to estimate the monthly average per-capita expenditure of district level in the Province of Bali, 2014 using direct survey estimates from the National Socio-Economic Survey (Susenas 2014) and auxiliary variables from the administrative record of village data (PODES 2014). The Fay-Herriot (FH) model (M1) as a benchmark and four P-Spline SAE model, i.e. the P-Spline SAE with spline term: x2 and x3 (M2), x3 and x4 (M3), x2, x3 and x4 (M4), and x1, x2, x3 and x4 (M5), are obtained. From RLRT, the spline term of M3, M4, and M5 are statistically significant. According to the parametric bootstrap of mean squared prediction error (MSPE) and coefficient of variation (CV), the M4 and M5 show a significant improvement with the CV values range from about 1%-6% compared to M1 with a range from about 4%-17%, shows that these two models more efficient. The M3 model shows the opposite performance even though has the smallest AIC value. More detail, the MSPE and CV produced by M4 are slightly better than M5 makes the M4 is the best model in this study.

Mixed Geographically and Temporally Weighted Regression Analysis of Percentage of Poverty in Java Island

Regression analysis is a method that can be used to analyze the linear relationship between explanatory variables with a response variable to analyze problems. The problem of poverty is one of the fundamental problems and is the center of government attention in every country especially developing countries like Indonesia, precisely in Java Island that have the largest number of poor people. In poverty data, there is spatial and temporal diversity. Besides, potential differences between regions in Indonesia allow for the diversity of data and can be locally and globally addressed so that the MGTWR method can be used. In this study using data on the percentage of poor people in regency/municipality in Java in 2012-2018. MGTWR modeling results show that the GDRP, literacy numbers, and EYS variables are global variables. Whereas the last variable of education was completed by elementary school, per capita expenditure, Raskin recipient households, residents of the 15-64 age group, and MYS were local variables. Based on the Bisquare kernel function, the MGTWR model is the best model to modeling the percentage of poor people in regency/municipality in Java with the smallest AIC value of 198.75 and PseudoR2 the biggest is 72.92%. In MGTWR modeling, 5 regional groups are formed based on factors that influence simultaneously.

PEMODELAN JUMLAH KASUS DEMAM BERDARAH DENGUE (DBD) DI JAWA TENGAH DENGAN GEOGRAPHICALLY WEIGHTED NEGATIVE BINOMIAL REGRESSION (GWNBR)

Dengue Hemorrhagic Fever (DHF) is one of the diseases with unsual occurrence in Central Java and spread throughout the regency/city. The number sufferers of this disease is still high because the mortality rate is still above the national target. Regarding the less handling of DHF spread, it is necessary to make a plan by identify the factors that allegedly affect that case. Characteristics of data the DHF cases is count data, so this research is carried out using poisson regression. If in poisson regression there is overdispersion, it can be overcome using negative binomial regression. Meanwhile to see the spatial effect, we can use the Geographically Weighted Negative Binomial Regression (GWNBR) method. GWNBR modeling uses a fixed exponential kernel for weighting function. GWNBR is better at modeling the number of DHF cases because it has the smallest AIC value than poisson regression and negative binomial regression. The results of research with poisson regression obtained three variables that have a significant effect on dengue cases. For negative binomial regression, two variables have a significant effect on DHF cases. While the GWNBR method obtained two groups of districts/cities based on significant variables. The variables affecting the number of DHF cases in all districts/cities in Central Java are the percentage of healthy houses, the percentage of clean water quality, and the ratio of medical personnel.Keywords: DHF, GWNBR, Poisson Regression, Binomial Negative Regression, Fixed Exponential Kernel

Estimating the time interval between transmission generations and the presymptomatic period by contact tracing surveillance data from 31 provinces in the mainland of China

The global pandemic of 2019 coronavirus disease (COVID-19) is a great assault to public health. Presymptomatic transmission cannot be controlled with measures designed for symptomatic persons, such as isolation. This study aimed to estimate the interval of the transmission generation (TG) and the presymptomatic period of COVID-19, and compare the fitting effects of TG and serial interval (SI) based on the SEIHR model incorporating the surveillance data of 3453 cases in 31 provinces. These data were allocated into three distributions and the value of AIC presented that the Weibull distribution fitted well. The mean of TG was 5.2 days (95% CI: 4.6–5.8). The mean of the presymptomatic period was 2.4 days (95% CI: 1.5–3.2). The dynamic model using TG as the generation time performed well. Eight provinces exhibited a basic reproduction number from 2.16 to 3.14. Measures should be taken to control presymptomatic transmission in the COVID-19 pandemic.

The Influence of Climate Factors on Cocoa Productivity in Sulawesi, 2019

Cocoa is one of the plantation commodities that has an important role in Indonesia's economic activity and is one of Indonesia's export commodities which is quite important as a source of foreign exchange and oil and gas. Sulawesi Island is one of the cocoa-producing islands in Indonesia. This study aims to determine a spatial regression model between the average cocoa productivity per month with the average drinking temperature per month, the average monthly rainfall and the average length of sunshine per month and the climatic factors that affect cocoa productivity in Sulawesi. The best model estimation uses the AIC value; the best model has the smallest AIC value. In this study, the SARMA spatial regression model is the best model with the specified criteria.

Intervention Model Analysis The Number of Domestic Passengers at Sultan Hasanuddin Airports

Domestic passengers are objects whose travel / flight transportation services only cover the domestic area. The increase or decrease in the number of domestic passengers is usually influenced by the occurrence of intervention. This research uses the intervention analysis. Intervention analysis is the time series analysis to model data that is determined by the presence of an intervention. Intervention analysis is one of the time series analysis to model data that are affected by the occurrence of a particular event in a short period of time, such as accidents, natural disasters, and promotions. This research is used to establish intervention model with pulse function of passengers of domestic Sultan Hasanuddin Airport. The result of the research were obtained the model Seasonal ARIMA .There were 6 intervention times during 2006 - 2018, by entering the intervention order b = 0, s = 0, and r = 1 based on the smallest AIC value is -303,66 with MAPE value is 6,1023.
 Keywords : Domestic passanger, Seasonal ARIMA, MAPE, Intervention analysis, pulse function

Glow-worm abundance declines with increasing proximity to artificial lighting

The glow-worm Lampyris noctiluca (Linnaeus, 1767) (Coleoptera: Lampyridae) is thought to be declining in the UK, partially due to Artificial Lighting At Night (ALAN). It is the aim of this study to investigate the influence of increasing proximity to artificial lighting on glowworm abundance. A survey of adult female glow-worms was carried out along 18 transects in two distinct areas of eastern England, Haddiscoe and Sizewell, where illumination by ALAN was known to be different, and the effect of distance to artificial lighting on glow-worm abundance was tested using generalised linear mixed models (GLMM). The largely-unlit Haddiscoe site had a higher abundance of glow-worms than the highly-lit Sizewell site. At the transect level, abundance of glow-worms declined significantly with increasing proximity to ALAN at both Haddiscoe and Sizewell, although the rate of decline varied between the two sites. The GLMM model with lowest AIC value and highest R 2 (98.10%) had a significant interaction effect between site and distance from artificial light. This suggests that even in landscapes with low ALAN, glow-worms are significantly more abundant at greater distances from sources of artificial illumination, and in very close proximity to ALAN the decline in abundance is precipitous. Given the strong reduction in glow-worm abundance due to ALAN, a threshold distance of 100 m is proposed to protect local populations; at closer proximity to ALAN the density of glowing females is seriously reduced. This recommended threshold will require further validation and testing across a wider geographical area.

Comparative Performance of the ARIMA, ARIMAX and SES Model for Estimating Reported Cases of Diabetes Mellitus in Anambra State, Nigeria

This study examined the performance of the ARIMA, ARIMAX and the Single Exponential Smoothing (SES) model for the estimation of diabetes cases in Anambra State with the following specific objectives: to fit the model to the data, to determine the best fit model for estimating diabetes mellitus cases and forecast for expected cases for period of five years. The secondary data used for the study is sourced from records of Anambra state Ministry of Health. The Akaike information criterion is adopted for assessing the performance of the models. The R-software is employed for the analysis of data. The results obtained showed that the data satisfied normality and stationarity requirements. The finding of the study showed that ARIMA model has least value of AIC of 1177.92, following the ARIMAX model with value of AIC=1542.25 and SEM recorded highest value of 1595.67. The findings further revealed that the ARIMA has the least values across the measures of accuracy. More so, five years predictions of the cases of diabetes mellitus were obtained using the models under study. From the results of the findings, ARIMA model proved to be best alternative for estimating reported cases of diabetes mellitus in Anambra state. Based on the findings, we recommend there is need for medical practitioners /health planners to create awareness and inform patients about the possible related risk factors of death through early diagnosis and intervention.

Comparative Performance of the ARIMA, ARIMAX and SES Model for Estimating Reported Cases of Diabetes Mellitus in Anambra State, Nigeria

This study examined the performance of the ARIMA, ARIMAX and the Single Exponential Smoothing (SES) model for the estimation of diabetes cases in Anambra State with the following specific objectives: to fit the model to the data, to determine the best fit model for estimating diabetes mellitus cases and forecast for expected cases for period of five years. The secondary data used for the study is sourced from records of Anambra state Ministry of Health. The Akaike information criterion is adopted for assessing the performance of the models. The R-software is employed for the analysis of data. The results obtained showed that the data satisfied normality and stationarity requirements. The finding of the study showed that ARIMA model has least value of AIC of 1177.92, following the ARIMAX model with value of AIC=1542.25 and SEM recorded highest value of 1595.67. The findings further revealed that the ARIMA has the least values across the measures of accuracy. More so, five years predictions of the cases of diabetes mellitus were obtained using the models under study. From the results of the findings, ARIMA model proved to be best alternative for estimating reported cases of diabetes mellitus in Anambra state. Based on the findings, we recommend there is need for medical practitioners /health planners to create awareness and inform patients about the possible related risk factors of death through early diagnosis and intervention.

PEMODELAN JUMLAH KASUS COVID-19 DI INDONESIA DENGAN PENDEKATAN REGRESI POISSON DAN REGRESI BINOMIAL NEGATIF

Pandemi COVID-19 telah merebak di seluruh penjuru dunia, termasuk Indonesia. Perlu diketahui berbagai faktor yang memengaruhi tersebarnya kasus COVID-19 di suatu wilayah sehingga baik pemerintah maupun masyarakat dapat melakukan upaya pencegahan dan penanggulanan agar pandemi ini tidak semakin menyebar. Penelitian ini bertujuan memodelkan jumlah kasus COVID-19 di Indonesia kemudian menentukan variabel-variabel yang berpengaruh signifikan terhadapnya. Model yang digunakan adalah regresi Poisson dan regresi Binomial Negatif. Kedua model dipilih karena merupakan model yang sering digunakan terhadap data cacah seperti jumlah kasus COVID-19. Kemudian dari kedua model akan dipilih model terbaik beserta variabel yang berpengaruh signifikan terhadap jumlah kasus COVID-19 di Indonesia. Unit analisis dalam penelitian ini adalah seluruh wilayah provinsi di Indonesia yang terdiri atas 34 provinsi. Variabel respon dalam penelitian ini adalah jumlah kasus kumulatif COVID-19 di Indonesia pada tanggal 9 April 2020 yang dihimpun dari Gugus Tugas Percepatan Penanganan COVID-19. Tanggal tersebut dipilih karena merupakan tanggal sebelum diberlakukannya kebijakan PSBB yang pertama kali di Indonesia dan penelitian ini tidak mengambil variabel intervensi pemerintah dalam memodelkan jumlah kumulatif COVID-19. Adapun variabel prediktor dalam penelitian ini adalah kepadatan penduduk, persentase penduduk lansia, persentase ruta dengan akses sanitasi tidak layak, dan persentase angka buta huruf penduduk usia 15 tahun keatas. Berdasarkan nilai AIC, regresi Binomial Negatif lebih baik digunakan untuk memodelkan jumlah kasus COVID-19 di Indonesia dibandingkan regresi Poisson.Pada model regresi Binomial Negatif, variabel kepadatan penduduk dan persentase penduduk lansia berpengaruh positif dan signifikan terhadap jumlah kasus COVID-19 di Indonesia.

Lasso Regression for Daily Rainfall Modeling at Citeko Station, Bogor, Indonesia

Abstract Rainfall is one of the climate components which is very important for an agrarian country with a tropical climate like Indonesia. There are so many variables that can affect the intensity of rainfall, including humidity, air temperature, speed and wind direction. However, not all of these variables can have a significant effect, therefore, we need a modeling technique that can select and shrink the predictor variables so as to make the model simpler. One of the techniques used is the Lasso (least absolute shrinkage and selection operator). The purpose of this study is to model rainfall intensity using lasso regression and determine which variable have the greatest influence on rainfall intensity. This paper also designs an application system that can be used to implement Lasso regression. The response variables in the form of daily rainfall at the Citeko observation station, Bogor Indonesia is assumed to have normal distribution. Rainfall intensity modeling involves 16 predictor variables, namely the maximum temperature, minimum temperature, average temperature, average humidity, sun exposure, maximum wind speed, wind direction when maximum wind speed and 8 other categorical variables are the most frequent wind direction. The resulting Lasso regression model successfully select and shrink the variables used to 9 variables only. In addition, based on the AIC and coefficient of determination indices for the model evaluation, the Lasso regression shows satisfying results rather than classical multiple linear regression. In Lasso regression, AIC value is 5901.20 where in classical model it is 5911.43. Also, Lasso regression coefficient of determination value is greater that is 22.73% while in classical model is 21.41%. From lasso regression model that has been obtained, it is known that the most frequent wind direction to the north is the variable that has the greatest influence on rainfall intensity. The Lasso regression application has been successfully created so that it can facilitate the user in uploading data, exploration of data and variables, Lasso regression analysis and report result.

Inequity in Education: Three-factor and Longitudinal Analysis

This study was conducted to find a more equitable education system for all students regardless of ethnicity, social status, gender, or even problems arising as a result of the Covid-19 pandemic. The effects of major factors that played a critical role on indicators of equity in education were analyzed. A data set from World Inequity Database on Education (WIDE) was used to gather information to estimate the inequity gap for different indicators and countries. The upper secondary completion rate was selected within this database to explore inequity in worldwide education and several stacked bar plots were drawn to describe the differences among demographics and incomes of countries. The learning achievement in reading(upper secondary) was chosen to explore inequity in education in Canada. A three-factor analysis model was built to study the effects of gender, location, wealth, and their interaction on learning achievement in reading. After model selection and model diagnosis, an additive model was chosen as the final model and it proved that there resided significant main effects. Also, a longitudinal model was built to explore whether observations of learning achievement in reading varied among the different years with four different levels as covariate variables. Three models related to repeated measures design were built, and the final model was chosen based on the AIC value. As a result, it provided a clear indication that the government should take responsibility and well-needed actions to eliminate inequity in education for many disparities that exist within comparisons between groups. Examples are: gender, rural or urban areas, and the social status of families.

Heterogeneity of growth and maturity of Larimichthys polyactis in the offshore waters of southern Zhejiang, China

The variations of life history traits have been observed for many fish species, which gains much concerns in the study of aquatic biology and ecology. In this study, the biological characteristics were explored for yellow croaker (Larimichthys polyactis) in the offshore waters of southern Zhejiang, based on 4920 individuals collected from 13 fishery-independent seasonal surveys from autumn 2015 to autumn 2018. Linear mixed effects models were used to estimate the growth, maturity characteristics, and their heterogeneity. The body length of yellow croaker samples ranged from 13 to 215 mm with the dominant body of 110 to 154 mm. The body weight ranged from 0.5 to 182.2 g, with the dominant body weight from 20 to 55 g. The results showed that the linear mixed effects models with random effects from season, gender, and year performed best for length-weight relationship, with the lowest AIC and RMSE values. The effects of season were much larger than those of genders and years. When the length exceeded 160 mm, the weight gain rate of yellow croaker was faster in spring and summer, lower in autumn and winter, while the male individuals gained more weight than the females with the same body length. Among 4841 individuals of specimens with gonadal data, the individuals at maturity Ⅱ stage occupied 50.4%, and the individuals at maturity stage contributed to 19.6%. The results from the best linear mixed effects model showed that season had the most significant influence on the maturity of yellow croaker. The 50% maturity length (L50%) was much lower in winter (124.6 mm) with no much difference between other seasons, indicating that yellow croaker matures earlier in winter. Our results indicated that linear mixed effect model could reflect the biological heterogeneity of yellow croaker conveniently and that the growth and maturity of yellow croaker had significantly sexual and temporal variations, which should be considered in the stock assessment and fishery management for yellow croaker.

Implementation of Using HMM-GA In Time Series Data

Abstract Some time series modeling methods have weaknesses, the static and dynamic information can not be consistently combined. Hidden Markov Model provides solutions to these problems. Hidden Markov Model (HMM) is an extension of the Markov chain where the state cannot be observed directly (hidden), but can only be observed through another set of observations. One of the problems in HMM is how to maximizing P(O|λ) where O is an observation and λ is a model parameter consists of transition matrices, emission matrices, and initial opportunity vectors which can be solved by the Baum-Welch algorithm. In practice, the Baum-Welch algorithm produces a model that is not optimal because this algorithm is very dependent on determining the initial parameters. To solve these problems, HMM will be combined with genetic algorithms (Hybrid GA-HMM). In general, based on AIC and BIC value, Hybrid GA-HMM is optimal than HMM.