Comparative Study of Machine Learning (ML) and Conventional Time Series Methodologies in Modelling the Exports Trade of Pakistan

• DOI: 10.59075/ijss.v2i2.232
• Publication Date: 2024-12-10T13:29:17Z
• Authors: Kinza Yousfani; Moiz Qureshi; Muhammad Daniyal; Muhammad Ismail

Export trade is a pivotal driver of economic growth and stability in any nation, including Pakistan. Accurate modelling export holds immense significance as it allows for informed decision-making strategic planning. By employing advanced techniques like machine learning alongside traditional time series models, we can gain deeper insights into the dynamics exports, anticipate trends, adapt policies strategies accordingly. This study focuses on comparative different models forecasting exports The dataset used this spans from 1972 to 2021, containing yearly data (% GDP) various MLP model, ELM classical ARIMA Exponential smoothing, model forecast that best met KPI criteria was chosen optimal candidate predicting behaviour data. outcomes concluded outperformed performance metrics, with MSE = 0.49, RMSE 0.70, MAE 0.53, MAPE 4.27. among all alternative Extreme Learning Machine Simple Smoothing. Our highlights effectiveness Multi-layer Perceptron (MLP) valuable tool exports. Its accuracy empowers policymakers make well-informed decisions, ensuring more sustainable prosperous future.