Analysing Optimising Time-Series Data for Demand Forecasting and Production Planning Using Bi-RNN

This research presents a solution to the problem of predicting demand and planning production. The proposed method utilizes Bidirectional Recurrent Neural Networks (Bi-RNN). Conventional approaches frequently face difficulties in comprehending the intricacies of time-series data, resulting in less-than-ideal decision-making in supply chain management. The suggested system incorporates several stages, namely data collection, preprocessing, feature selection, implementation of Bi-RNN architecture, and integration with production planning systems. Our Bi-RNN model outperforms existing methods such as ARIMA, Simple Exponential Smoothing, LSTM, and Random Forest in terms of performance measures. Our model demonstrates superior performance compared to baselines, as evidenced by its Mean Absolute Error (MAE) of 3.4, Root Mean Squared Error (RMSE) of 4.2, Mean Absolute Percentage Error (MAPE) of 2.8%, (R Square) value of 0.95, Mean Squared Error (MSE) of 17.64, and Forecast Bias (FB) of 0.1. This research enhances the area by presenting a strong framework for predicting demand and planning production. It provides more precise forecasts and allows for smooth interaction with production systems. The proposed Bi-RNN technique has potential benefits for optimizing inventories, reducing costs, and enhancing customer satisfaction in supply chain operations.