Forecasting halal tourism economic impact on regional resilience

Abstract

Purpose – This study aims to forecast the economic contribution of halal tourism to the Gross Regional Domestic Product (GRDP) in the Special Region of Yogyakarta (DIY) and assess its role in regional economic resilience using recent Central Bureau of Statistics (Badan Pusat Statistik, BPS) data and advanced machine learning models.
Methodology – Quarterly GRDP data (2019–2025) were interpolated to monthly data and integrated with BPS January 2026 statistics (international tourists, domestic tourists, hotel occupancy). Hybrid modeling: XGBoost machine learning model for feature importance and long short-term memory (LSTM) neural networks with 12-month historical patterns for forecasting. The features include domestic tourists to Sleman, prior-month hotel occupancy, seasonality indicators, and halal demand proxies (ASEAN/Malaysia share).
Findings – XGBoost achieves a mean absolute percentage error (MAPE) of 3.98%, with domestic tourists to Sleman (75.8%) and prior-month hotel occupancy (10.8%) as dominant drivers. LSTM achieves a MAPE of 6.13% on the test set (2024–2026), capturing seasonality and low-season decline in January 2026 (star-rated hotel occupancy 50.77%, domestic tourists down 8.84% month-on-month). Forecasts to December 2026 show GRDP share stabilizing around 8.5–9%, resilient post-pandemic but vulnerable to seasonal shocks.
Implications – Halal tourism bolsters DIY resilience via domestic Muslim demand (domestic tourists in urban areas of Sleman). Policies should target off-season halal certification for hotels to flatten volatility and sustain the GRDP share.
Originality – This is the first study to integrate BPS January 2026 granular data with hybrid machine learning forecasting for halal tourism in DIY, validating urban halal hubs as a key resilience factor.