<![CDATA[The growth of the digital economy encourages the development of subscription-based business models in business management applications, including in the laundry sector. This study aims to analyze the historical data patterns of the Matik Laundry Application revenue, compare the accuracy of three time series forecasting methods, determine the best model, and produce revenue projections for the 2026–2028 period. The data used are monthly data on subscription revenue, the number of new users, and the number of subscription users for the period January 2023 to December 2025 as many as 36 observations. The methods applied include Autoregressive Integrated Moving Average (ARIMA), ARIMA with Exogenous Variables (ARIMAX), and Exponential Smoothing Holt-Winters. The data is divided into 30 months of training data and 6 months of test data. The stationarity test using Augmented Dickey-Fuller (ADF) shows that the data is not stationary at the original and stationary levels after first-order differentiation. The analysis of ACF and PACF yielded an optimal order of ARIMA(2,1,0). The results of the evaluation based on the Mean Absolute Percentage Error (MAPE) showed that Exponential Smoothing Holt-Winters was the best model with a MAPE of 5.98% (very accurate), followed by ARIMAX(2,1,0) with a MAPE of 8.19%, and ARIMA(2,1,0) with a MAPE of 14.38%...]]>
Copyrights © 2026
