ANN-Excel: Artificial neural network framework in excel

초록

In this paper, we propose an Artificial Neural Network in Excel (ANN-Excel) software. This software implements neural network-based prediction feature by utilizing Excel, which is easy to handle data in a spreadsheet and easy to perform using Visual Basic for Applications (VBA) code. The software has been implemented adopting feed-forward multilayer perceptron, error back-propagation algorithm, momentum process, and data scaling, enabling users to utilize machine learning technique directly in Excel without prior knowledge of mechanism. Also, the performance of ANN-Excel was compared with WEKA and Keras in Python, via Sum of Squared Error, Root Mean Squared Error(RMSE), and R-Square metrics. The results show that ANN-Excel achieves comparable or superior performance to WEKA and Keras across benchmark datasets, while remaining fully operable within Excel. This work highlights ANN-Excel as an accessible and reproducible tool for integrating neural networks into spreadsheet-based analysis.

키워드