With the development of IT, immediate prediction through real-time data is becoming important in various fields such as finance and manufacturing. However, existing models have limitations in reflecting the autocorrelation of real-time data and suffer...
With the development of IT, immediate prediction through real-time data is becoming important in various fields such as finance and manufacturing. However, existing models have limitations in reflecting the autocorrelation of real-time data and suffer from inefficiency, requiring the entire model to be retrained when new data arrives. To solve this problem, this study proposes a Online regression model OLDV including lagged dependent variables (LDV) and ORLDV applying ridge
normalization. To validate the model's superiority, we compared its predictive performance against existing models like GLM, RLS, and LSTM under conditions with extremely limited training data. As a result of the experiment, the proposed model significantly reduces errors compared to existing models and demonstrates superior performance. This study is significant for presenting a new methodology that simultaneously enhances the accuracy and practicality of online predictions.