This study is to propose a framework for sustainable supplier evaluation and optimal allocation of orders for multi-product and multi-period. In order to forecast demand, the study employed a machine learning algorithm. Concurrently, an integrated Mathematical programming model, encompassing four crucial sustainability factors, is formulated. This model effectively allocates purchase orders to suppliers based on the outcomes of the evaluation process. The model also accommodates the distribution of orders across various time frames and diverse product categories. Subsequently, the effectiveness and robustness of the proposed model are assessed through sensitivity analysis via a real-world case study implemented within an authentic supply chain setting.