Optimization of MSE Accuracy Value Measurement Applying False Alarm Rate in Forecasting on Fuzzy Time Series based on Percentage Change

Time series is a method of forecasting in which it is carried out on the previous data to obtain future data. This study conducted forecasting on the price of gold, in which the process is divided into two patterns. In the first pattern, forecasting used change-based time series with the world economic instability such as the Covid-19 pandemic, and the second pattern is that using time series based on changes with the stability of the world economy. In forecasting, the test level is measuring forecasting accuracy. It can be done by MSE, which is still so easy that optimization is carried out in measuring accuracy by modifying MSE using a false alarm rate in the form of a formula to find the difference. The first pattern process went well and got an accuracy value calculated with the MSE of 0.0538, and the modified MSE measurement technique used the False Alarm Rate of 0.0119. In the second pattern, it also went well and got an MSE value of 0.1508, and the modified MSE measurement technique used a False Alarm Rate of 0.446. Forecasting using change-based time series could affect the amount of accuracy based on the amount of training data so that the more training data, the smaller the accuracy value was obtained, while the less training data, the greater the accuracy value.