ERFLSTM: Enhanced regularization function in LSTM to assess feature importance

Feature importance is an important technique of solving the pattern of prediction and the regression tasks. There is no need to train the model with all the features in the dataset. Minimizing the features reduces the load during the prediction phase. This study proposes an improved regularisation technique to reduce the error and solve the over fitting issues to identify the relative importance of features. The more important feature the more weight and connected to the corresponding input neurons during the training of the deep learning model. We compared the proposed with seven state of the art feature importance techniques on different data sets. The proposed regularisation function outperforms the other state of the art feature importance techniques. The Proposed Approach achieves superior performance metrics with an accuracy of 93.78%.This method stands out due to its ability to deliver highly accurate, precise, and comprehensive results, reflecting its advanced regularization and dynamic updates. Its exceptional performance underscores its effectiveness in refining feature importance and enhancing model interpretability.