Comparison of ARDL and artificial neural networks models for foreign direct investment prediction

Abstract EN

This paper examined two common approaches, the ARDL model econometric technique and the artificial intelligence processes known as the artificial neural network model, which were applied to the forecasting process of FDI flows in Algeria from January 1990 to December 2019.

According to our empirical study, The time series were tested and found to be stationary in the first order, which would require the use of the cointegration test, and using the Bounds test, it was found that there was a long-run relationship between the variables.

The ARDL model is estimated to perform the prediction.

Thereafter, the optimal neural network was designed, consisting of three nodes in the input layer and 12 nodes in the hidden layer.

According to the Training algorithm, optimal weights and biases were determined, so that we could finally implement the prediction process, according to the MSE and RMSE prediction standards and found that the ANN model had better on the ARDL model in forecasting FDI flows in Algeria.