A novel hybrid holt integrated moving average (HIMA) model for enhanced forecast accuracy in trend data series.

Holt's method is one of the most popular forecasting techniques for time series, particularly with trend variations. Unfortunately, due to limitations of Holt's method, such as sensitive parameter selection, the linearity assumption requirement for such a model can lead to overestimation or underestimation, especially for different trend variations. This study aims to introduce a hybrid Holt's method by integrating the traditional Holt's method and the Moving Average (MA) in Box-Jenkins methodology called Holt Integrated Moving Average (HIMA) to improve forecast accuracy for different trend variations. Eighteen simulated datasets, with six different sample sizes, such as n=50, 100, 150, 500, 1000, 2500 and three different trend variations: linear, cubic, and quadratic were used to evaluate the model performance. Besides that, two real datasets, which Consumer Price Index (CPI) and PETRONAS share price, were used to validate the model performance.