V3I5P65

Denton Proportional Disaggregation and SARIMA Modeling for E-Commerce Sales Forecasting Under Data Scarcity: A Case Study of TiKi Vietnam 

Phan Thanh Tung^1*    

Abstract

Amid the rapid expansion of Vietnam’s e-commerce sector, TiKi faces a severe growth slowdown, contrasting with the general market trajectory. Academic and managerial insights into this platform are heavily constrained by data scarcity, as its Gross Merchandise Volume (GMV) data is exclusively available at a low (quarterly) frequency. To address this limitation, this study develops an integrated temporal disaggregation and econometric forecasting framework. First, the quarterly GMV of TiKi is disaggregated into a monthly series using the Denton Proportional method mapped against Google Trends search volume indicators. Subsequently, a Seasonal Autoregressive Integrated Moving Average (SARIMA) model is formulated to capture complex historical trends and seasonal fluctuations. The empirical results demonstrate that the optimal SARIMA framework achieves high predictive accuracy and satisfies all rigorous residual diagnostics. Methodologically, this study validates the synergy between online search behavior data and traditional time-series models, providing a robust, low-cost forecasting solution for strategic decision-making under severe data constraints.

Keywords:

E-commerce, Tiki, Denton Proportional, Google Trends, Sales Forecasting, SARIMA.