Improve forecast accuracy by using repeated time series cross validation (RTS-CV)): a case study of Digi share price

Forecasting is the entire process of creating the essential techniques to produce future values that can then be utilised as inputs suited to the aims and objectives of the company (Alias, L. 2011). Based on data from prior experience, forecasters can make precise choices for the near future. The goal of this study is to make predictions about the share price of Digi Telecommunications Sdn. Bhd. (DI GI) on the stock market. Azlan Abdul Aziz (2021) acted as the representative for this approach. Given that there are many instances of erroneous predicting, this strategy is an improvement over the prior one. Inaccurate predictions of future values will result in poor decision-making and, even worse, might cause investors and stockholders to become fearful. The models that produce the lowest error measures were collected and compared to decide the most excellent predictions for this study. The next step is to distinguish the most excellent performance from the five models utilized in this consideration. This study was conducted to predict Digi's share price on a daily, weekly, and monthly basis for data high and low only from May 9, 2006 to May 2, 2023. Outcomes for univariate time-series investigation models, such as the Naive, Mean, Single Exponential Smoothing (SES), Holt's, and ARIMA models, were analyzed. Five sets of data splits (high and low) were used for each model to ensure the accuracy of the predicted values. Moreover, the smallest error for instance RMSE, MAE, MAPE, and MASE, are vital in deciding the demonstrative execution, with lower values showing more productive prescient model.