ساختار ژنتیک، ازدواج های فامیلی و توسعه اقتصادی: هم انباشتگی پنل و تجزیه و تحلیل شبکه های عصبی هم انباشتگی پانل
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|6974||2011||11 صفحه PDF||سفارش دهید||11000 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Expert Systems with Applications, Volume 38, Issue 5, May 2011, Pages 6153–6163
Consanguineous marriages and their effects on human beings in light of biological effects of genetic sicknesses are discussed in many studies. Among many, the likelihood of sicknesses such as phenylketonuria, thalassemia, Landsteiner–Fanconi–Anderson’s syndrome, hemophilia and many neuro system anomalies increase drastically in countries with consanguineous marriage practices resulting in increasing economic costs. In the study, we aimed to analyze the effects of consanguineous marriage and its effect on economic growth and development. We also analyzed infant mortality in these countries in light of consanguineous marriages and economic development. In the study, Panel Cointegration specifications are integrated into Neural Network models known with their strong generalization properties. The study focuses the econometric analyses, where the Panel Cointegration Neural Network Model is investigated and compared to the Panel Cointegration Model. According to MSE, MAE and RMSE error criteria and Diebold Mariano tests of equal forecast accuracy, the results suggest strong advantages of Panel Cointegration MLP models compared to Panel Cointegration models used in regression analysis.
We will investigate the effects of consanguine marriage and genetic structure on economic growth and development. The study has two objectives. These are, to look at economic development from a different framework and to collate the Panel Cointegration and neural network literature around the axis of cointegration. The first part contains the econometric theory, where the Panel Cointegration MLP Model is investigated and compared to the Panel Cointegration Model. In the study, as panel structure is merged with neural network structure, the panel cointegration structure is aimed to be augmented with the neural network. Multi Layer Perceptron (MLP) are combined with panel structure and the peformances of the discussed models, Panel Cointegration MLP Model (PCNN) will be obtained to be compared with Panel Johansen Cointegration Model. The second part of the study focuses on economic development and growth, consanguine marriages, infant mortality and genetic structure relations in theoretical framework.
نتیجه گیری انگلیسی
In the study, the effects of consanguineous marriage and infant mortality rates are investigated with panel cointegration regression and panel cointegration neural network models. The outcome of the study is 2-fold. First, in light of the generalization and learning characteristics of neural networks, panel cointegration methodology is extended to neural networks models for panels with country and time dimensions. One important difference of the study is that, though neural network models are nonparametric, the study provides Panel-MLP and Panel-VEC-MLP models in parametric form. All Panel Cointegration MLP models have linear identity activation functions in the hidden layer and output layers. For Panel-VEC models, we observed that increases in consanguineous marriage and infant mortality rates have an inverse effect on GDP per capita for all country groups analyzed. By moving from Group 1 to Group 4; from the country group with lowest consanguineous marriage to the group with the highest similar results are achieved within the models. Countries with higher rates of consanguineous marriages and infant mortality levels are more likely to experience comparatively lower levels of economic development. The fit of the models are evaluated with MSE, MAE and RMSE error statistics. The forecast accuracy is evaluated both between Panel-VEC-MLP and Panel VEC models, in addition to the long run models. The results suggest significant improvement in modeling error correction if Panel Cointegration models are augmented with neural networks models.