Accurate forecasting of short-length multivariate time series data is essential for decision making in domains where data scarcity, non-stationarity, and complex dependencies create significant challenges. This study proposes an Artificial Bee Colony-optimized Long Short-Term Memory (ABC-LSTM) model to address these challenges and support Sustainable Development Goal 8 (SDG-8), which focuses on sustainable economic growth, decent work, and economic resilience. The model leverages critical features, including adjusted growth, modern slavery, unemployment rates, and human rights, to enhance prediction capabilities. To overcome data limitations, data augmentation techniques, such as upsampling, are applied, ensuring robust performance on limited datasets. The model achieved exceptional accuracy, with MSE of 0.1140, RMSE of 0.2119, MAE of 0.1672, and R2 of 0.4867, demonstrating its effectiveness in forecasting short-length time series. Predictions for SDG-8 progress from 2022 to 2026 indicate stable improvement, ranging from 69.64 to 70.22. These results underline the reliability and ability of the model to guide data-driven policymaking for sustainable development objectives.