Comparison of Novel Long Short-Term Memory Over Recurrent Neural Networks for Predicting the Melting Pattern of Ice in Antarctic Sea Based on Accuracy

Revolutionary technique that blends revolutionary Novel Long Short-Term Memory (NLSTM) with Recurrent Neural Network (RNN), this research seeks to revolutionize the prediction of ice melting. This research paper's dataset, which has 212 rows and 3 columns (attributes), was obtained from Kaggle.com. This has to do with Antarctic mass, year, and 1-sigma uncertainty. The clincalc web application was utilized to determine a total sample size of 92 for the paper. Gpower sets a value of α=0.05 and power=0.80 to find a sample size of 20 for each group. Based on the data obtained from the two groups NLSTM has a much greater accuracy (98.64%) as compared to that of RNN accuracy (96.78%). Statistical significance was noted between NLSTM and RNN with p = 0.002 (Independent Sample T-Test, p < 0.05). In summary, the research indicates that the NLSTM outperforms the RNN in terms of accuracy in which the Climate in Change Factor dataset is used.