A Robust Demand Forecasting Model via N-Valued Neutrosophic Trapezoidal Numbers with Significant Feature Selection for Enhanced Business Intelligence

Nowadays, Business Intelligence (BI) plays a pivotal role in implementing effective strategies and informed decision-making based on data. It is an essential decision support system that enables organizations to analyze data and overall business processes. Demand forecasting is a major decision-making task for enterprises and is a well-established branch of predictive analytics that has gained significant attention over time. Currently, Machine Learning (ML) is widely employed as a key tool, alongside other technologies, to apply BI concepts in demand forecasting for various business domains. This paper presents a robust Demand Forecasting Model with N-Valued Neutrosophic Trapezoidal Numbers and Feature Selection (DFM-NVNTNFS) model for intelligent business applications. The main objective of the proposed DFM-NVNTNFS method is to enhance BI through an advanced technique-based forecasting framework that improves prediction accuracy and decision-making capability. Initially, the Z-score method is employed in the data preprocessing phase to transform raw business data into a structured format. For feature selection, Recursive Feature Elimination with Cross-Validation (RFE-CV) is used to identify the most relevant features from the dataset. Subsequently, the proposed DFM-NVNTNFS model incorporates N-Valued Neutrosophic Trapezoidal Numbers (N-VNTN) for demand prediction modeling. The effectiveness of the proposed DFM-NVNTNFS approach is validated through comprehensive experimental analysis using a benchmark dataset. The comparative results demonstrate that the proposed method achieves improved performance and scalability across different evaluation aspects compared with existing approaches.

Hali tarjima qilinmagan