ARIMA ANNالتنبؤ بتدبدبات اسعار الصرف دراسة قياسية لحالة الجزائر نمودج الشبكات العصبية خلال الفترة 1960-2018

Abstract

تهدف هذه الدراسة إلى قياس تقلبات أو تذبذب سعر الصرف الدولار مقابل الدينار في الجزائر خلال الفترة ( 1960-2018 )، وكيفية نمذجتها باستخدام نموذج ARIMA وANNوذلك من اجل التنبؤ يقيمها، ولتحقيق هذا الهدف يتم تقسيم الدراسة إلى فصلين، حيث خصص الفصل الأول للجانب النظري والذي يحتوي على كل من مفاهیم سعر الصرف و والشبكات العصبية وايضا الدراسات السابقة الدراسات السابقة، اما الفصل الثاني تضمن الجانب التطبيقي للدراسة وذلك باستعمال التحليل الإحصائي في تقدير النموذج وما يتطلب من اختبارات، وبعد اختبار الفرضيات إحصائيا باستخدام تحليل الانحدار الخطي و الذي يعد أحد الأساليب الإحصائية الكمية، حيث تم توصل إلى وجود تغيرات مترافقة بين سعر الصرف الرسمي و الموازي مع توسع الفجوة بينها وبالتالي نتائج الاختبارات الإحصائية تؤكد حسن تخصیص أو اختيار النموذج، أي أنه صالح لنمذجة أسعار صرف الدولار مقابل الدينار .

The results of the comparison between ARIMA and ANN show us on several levels: First: On the conceptual level The ARIMA method, which is the oldest method since its inception, was founded by two scientists Box and Jenkins, which benefited from some physical and mathematical concepts, especially those related to simple and partial self-correlation functions. Relatively and more complicated, work is still being done to develop them, especially in software that gives ready results. Second: At the level of results and expectations These two methods are similar in working on obtaining some results, such as graphical curves, correlations, reliance on residuals, errors, etc. and differ in other details, and as for the forecasts or expectations that we obtained in our case, it appears that the exchange rate using the ANN method in Algeria is that this price will stabilize in the coming years (and this maybe Looking at the plans of the Algerian government that is trying to target it in the future in order to restore some prestige to it in exchange for other foreign currencies) and for the ARIMA method, it appears that it will rise in the coming years, but with a small tendency (and this is perhaps because the deterioration of the Algerian dinar .cannot be curbed in the near future, especially with the economic conditions in Algeria. Which cannot stop its deterioration, at least in the near future.) The one who contemplates both the prediction results of the two methods cannot be certain which is the correct one or which will be achieved because we are first in the state of uncertainty and the unknown, and both methods have been and are being developed each time with their support with tests and other tools that correct some of the imbalances in their results. Data for a specific phenomenon, while the opposite may happen in another.