The Comparison of Missing Value Imputation for Price Index Forecasting Based on ARIMA Model

Investments in the modern era exhibit a greater range of diversity. In the contemporary period, advancements in technology have facilitated investors' access to many investment opportunities. Stock market investments provide a greater probability of generating profits compared to other investing options. Consequently, several investigations have been carried out to determine methods for predicting stock price indices. However, due to the availability of stock price index information only on days when the stock market is operational, there is a lack of data on some days. This study introduces a technique for incorporating price index data into forecasting models by using the FFILL, BFILL, and interpolate approaches. Subsequently, the newly acquired data is used for forecasting using the ARIMA model. The experiment yielded findings indicating that prioritizing data input enhances the accuracy of prediction outcomes.