Volume 19, Issue 1, Winter and Spring 2022

Support Vector Regression (SVR) model

Noora Hawa; Salih Muayad

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2022, Volume 19, Issue 1, Pages 1-16
DOI: 10.33899/iqjoss.2022.174327

In this paper, the Support Vector Regression (SVR) model was used, which is defined as an algorithm or a linear model used to predict a specific model. The performance efficiency of the SVR method depends on the selection of its hyperparameters. In this paper, the SVR method was used with the Strawberry Algorithm, which is the proposed algorithm to obtain the best combination of hyperparameters.
The Root Mean Squares Error (RMSE) criterion was used to compare the results obtained using the proposed algorithm with some common algorithms, namely, Grid Search, Genetic Algorithm, Particle swarm optimization, and an annealing algorithm (Simulated Annealing algorithm. Three methods of selection were also used in the strawberry algorithm, roulette wheel selection, elite selection, and roulette wheel with the elite selection method together. The performance of the algorithm was tested through experimental and real data. The results showed that the strawberry algorithm was superior to the common algorithms in choosing the best combination of hyperparameters. The results also showed that the method of choosing the roulette wheel is the best method that gave good results compared to other methods on the experimental and applied sides.

Using Wavelet Shrinkage in the Cox Proportional Hazards Regression model (simulation study)

taha ali; Jwana Rostam Qadir

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2022, Volume 19, Issue 1, Pages 17-29
DOI: 10.33899/iqjoss.2022.174328

The proposed method in this paper dealt with the problem of data contamination in the Cox Proportional Hazards Regression model (CPHRM) by using Wavelet Shrinkage to de-noise data, calculating the discrete wavelet transformation coefficients for wavelets (Symlets and Daubechies), and thresholding methods (Universal, Minimax, and SURE), as well as thresholding rules (Soft and Hard). A software in the MATLAB language built for this propose will compare the proposed and classical method using simulation and real data. All the proposed methods have better efficiency than the classical method in estimating the Cox Proportional hazards model depending on both average of Akaike and Bayesian information criterion.
 Keywords: Cox PH model, Wavelet Shrinkage, thresholding rules.

Comparison of prediction using Matching Pattern and state space models

heyam hayawi; najlaa saad

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2022, Volume 19, Issue 1, Pages 30-37
DOI: 10.33899/iqjoss.2022.174329

Predicting future behavior is one of the important topics in statistical sciences due to the need for it in different areas of life, and most countries rely on their development programs on advanced scientific foundations and methods in order to reach more effective results. This research deals with a comparison of the accuracy of time series prediction using state space models and the matching patterns method of Singh (2001) algorithm by applying to real data, which are economic observations that were previously addressed by the researchers Box and Jenkins (1976). Where the inputs represent the leading indicator and the outputs represent sales, and the importance of this research is represented in Knowing the most accurate method for predicting time series. The MATLAB program has been used to access the results of the research. The most important results of the research are that the state space model is more accurate in forecasting than the matching patterns in the studied data because it has the lowest values ​​of the test criteria of prediction accuracy results.

Spatial Prediction of Real Sulfur Data Using the Ordinary Kriging Technique and Lognormal Kriging

Najlaa Sadeek

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2022, Volume 19, Issue 1, Pages 38-45
DOI: 10.33899/iqjoss.2022.174330

This research deals with the spatial prediction process in order to obtain the optimal prediction when the data are distributed normally. In this paper, we used the ordinary kriging technique and the lognormal kriging after taking the logarithm of the original sulfur data. We used the variogram function in this research to get the best model for the covariance function. The aim of this research is to evaluate the normal kriging and the lognormal kriging and find outliers. The data adopted in this work are from the hydrogeological study of Mosul Governorate/Iraq. Through the results, it was found that the errors in the estimated value are very important for the variance of the estimator, which appears to be very small. As well as through the results that were supported by graphs, we note that the lognormal kriging has more effect than the ordinary Kriging technique under the prediction process, from During the implementation of the error tests which seemed to be very small and which support the predictive values ​​of the spatial sulfur data, the MATLAB programming language was used to obtain the practical results.

Shrinkage estimators in inverse Gaussian regression model: Subject review

Farah Abd ulghani; رفل الحمدانی

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2022, Volume 19, Issue 1, Pages 46-53
DOI: 10.33899/iqjoss.2022.174331

The presence of the high correlation among predictors in regression modeling has undesirable effects on the regression estimating. There are several available biased methods to overcome this issue. The inverse Gaussian regression model (IGRM) is a special model from the generalized linear models. The IGRM is a well-known model in research application when the response variable under the study is skewed data. Numerous biased estimators for overcoming the multicollinearity in IGRM have been proposed in the literature using different theories. An overview of recent biased methods for IGRM is provided. A comparison among these biased estimators allows us to gain an insight into their performance.

Generalized ratio-cum-product type exponential estimation of the population mean in median ranked set sampling

Rikan .AL_Rahman; Saja Mohammad

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2022, Volume 19, Issue 1, Pages 54-66
DOI: 10.33899/iqjoss.2022.174332

This study presents a proposal to estimate the finite population's mean of the main variable by median ranked set sampling MRSS through the generalized ratio-cum-product type exponential estimator. The relative bias PRB, mean squared error Mse, and percentage relative efficiencies PRE of the proposed estimator is obtained to the first degree of approximation. The proposed estimator are more efficient than the usual unbiased estimator, ratio, product type, and other estimators. Finally, the estimators' abilities are evaluated through simulations, showing that the proposed estimator is more efficient than several other estimators.

Identification of Transformation Function Models for OPEC Crude Oil Prices

hashim huseen; najlaa saad

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2022, Volume 19, Issue 1, Pages 67-75
DOI: 10.33899/iqjoss.2022.174333

The transformation function model is one of the basic concepts in time series as it deals with multivariate time series. As for the design of this model, it depends on the data available in the time series and on other information in the series. Therefore, the representation of the transformation function model depends on the representation of data and the accuracy of the available information. and use this information in modeling. The research aims to identification the transformation function model of the monthly time series of crude oil barrel prices of the Organization of Petroleum Exporting Countries (OPEC) in US dollars as a series of outputs and the price of Brent oil as a series of inputs during the time period from (2005) to (2019). The transformation function model with the order (s,r,d,pn,qn)=(2,2,0,2,3) is the best for representing the data and the mean error criterion was used to know the prediction accuracy of the estimated transformation function model for nine months and its value was ME=-0.00851  negative That is, most of the errors are negative, which is evidence that the approved prediction gives optimistic results.

Detection of outliers in the linear regression model with application to well water pollution data on the outskirts of the city of Mosul

Saja M. Ismail; Safwan Nathem Rashed

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2022, Volume 19, Issue 1, Pages 76-84
DOI: 10.33899/iqjoss.2022.174334

The research idea is concerned with identifying the effect of outliers on the parameters of the multiple linear regression analysis models. Where the outliers values present in the data are detected and diagnosed if they are in the independent or the dependent variable, which causes an impact on the estimation of the parameters of the studied model.  The extreme data types and methods of processing them have been identified to obtain a better model with high efficiency or reduce the impact of These values on the model; the MSE standard was developed to compare treatment methods and was applied to real data taken from the Dams and Water Resources Research Center, University of Mosul. As suggested by (شاکر،2009) is the best in detection among the methods that have been used. 

Treatment of time series instability - review article-

Alla abd alsatar; Nada Nazar Alobaidi; Zainab Tawfeq

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2022, Volume 19, Issue 1, Pages 85-93
DOI: 10.33899/iqjoss.2022.174335

the    time series is a problem in econometric analysis as the statistical properties of series analysis are lost when using unstable time series. The research aims to present several methods for dealing with stability, including (Box Jenkins, Exponential Smoothing, Double Exponential Smoothing, Exponential Smoothing Moving Averages, Fuzzy, Neural Network) and to compare the methods presented through diagnosing ARIMA models after achieving stability and choosing the best method that corresponds to the lowest values ​​of the criteria Statistics (MSE, AIC, BIC). The above-mentioned methods have been applied to daily data for the year 2020 to generate electricity from water coming from the Tigris River, and it was concluded that the (Fuzzy) method is the best for treating stability compared to other methods for having the ARIMA model (0,1,3) corresponding to the lowest values ​​of the criteria Statistics (MSE=0.572, AIC=-196.4536, BIC=-0.6931).