Galoitica: Journal of Mathematical Structures and Applications
Journal DOI
2834-5568ISSN (Online)
This work is informed by a time analysis procedure to investigate evaluation of the study variables from 2005 to 2023, employing of time- series models to reveal the relationship between governance and corruption through application of EViews13. The corruption perceptions index is considered as dependent variable, government effectiveness and rule of law are assumed independent variables, whereas the corruption growth rate is control variable. To describe model variables, time series regression models are employed subsequently, the graphical analysis and Augmented Dickey-Fuller (ADF) test are applied to know time series stationarity. Also, the two tests are utilized to examine degree of integration, nature of stochastic approach for variables, in case dependent or independent, over study time. This is done to ensure that the time series of the variable is stationary, meaning that it has a constant mean across its values, a constant variance, and no trend. Moreover, the tests conducted at first and second differences to assess stationarity at a certain difference, ensuring achievement stationary series adequate for modeling. Furthermore, robustness and validity of the estimated model are evaluated by autocorrelation tests and diagnostic tests of residuals. The findings show that the graphical method applied to examine stationarity is not highly precise as it appears severe oscillations in all variables under study. This involves conducting a stationarity test for the time series in according to equation of constant and trend by using the ADF test under the first and second differences. At the second difference I(2) , all variables are stationary as results show and the model can estimate a statistically significance model, with a coefficient (R2 =96.3%), referring that model can explain the changes in corruption perceptions and has a high explanatory potential. Also value of F- statistic is high, reaching (183.3882). The time-series results further demonstrate presence of a statistically significant relationship between corruption and governance. It is revealed that advancement of government effectiveness exhibited statistical significance at 99% confidence level, and it is considered one of crucial independent variables in decreasing growth of corruption. The results show that a one-unit increase in government effectiveness results an increase corruption reduction by approximately (53.43) points. Concerning acceleration of rule of law, it has major influence over reducing the corruption. The outcomes state a one-unit increase of advancing rule of law deceleration of corruption acceleration by approximately (26.42) points, at a 99% confidence level. Furthermore, value of the constant of estimated equation is ( -0.07) at the 95% confidence level, suggests that there is an escalation of corruption because of absence of governance indicators, which driven by endogenous factors of Iraqi society. This characterizes the preceding value of corruption before adopting of governance regulations. In the model residuals, lack of autocorrelation is characterized by the Durbin-Watson, while value of S.E. of regression reveals a high level of estimation precision. Hence, it can be deduced that governance stringent procedures contribute to decreasing corruption rate over the study period.
Read MoreDoi: https://doi.org/10.54216/GJMSA.130102
Vol. 13 Issue. 1 PP. 10-22, (2026)
Breast cancer is a common type of cancers and the main reason of increased death of women universally. Recently, ML methods have become important in varying fields, such as Logistic Regression, Elastic Net Logistic, Decision Tree, Random Forest, Boosting, Naive Bayes and K Nearest Neighbor. The aim of the current study is to know and predict the type of cancerous tumor whether it is benign or malignant. These above techniques are expected to be helpful. Breast tumor type diagnosis using numerous performance metrics i.e. accuracy, classification error, sensitivity and specificity, both certified and trained models were assessed. The models were developed to determine which model would provide the best performance and comparisons were done. A separate data set from the one used to create the models was utilized to confirm every model. According to the analysis, the findings showed that elastic net logistic model had the highest performance in accurate classification rate (accuracy), classification error and sensitivity. Making it the best classifier for predicting the kind of breast cancer among all other models, privacy and it was also distinguished by reduce the high dimensionality and multicollinearity problems.
Read MoreDoi: https://doi.org/10.54216/GJMSA.130101
Vol. 13 Issue. 1 PP. 01-09, (2026)