Evidence record · UDINUS · SDGs
- Unit
- Rektorat UDINUS
- Period
- 2022
- Goals
- Record
- 0a70947d-05d3-44fb-9495-d184fc02a45d
- Published
- 2026-10-30
Universitas Dian Nuswantoro conducted a study to determine the effect of COVID-19 on movements in the Jakarta Composite Index (JCI) stock price by testing combinations of input variables derived from JCI closing-price data. The activity used a neural network method to identify the combination of input variables that produced the best Root Mean Square Error (RMSE). The resulting RMSE was then compared between grid search parameter optimization and evolutionary parameter optimization.
The data were obtained from the JCI Historical Data page on Yahoo Finance for the period from 12/11/2019 to 12/30/2021 during the COVID-19 pandemic. The dataset contained 509 records. Closing-price data were used as the input variable and target. Data preprocessing consisted of data cleansing, filtering, and windowing extending to seven days before the relevant closing price. The analysis produced an RMSE of 0.104 for five days before Close t (P=5), with a training cycle of 9000 and momentum of 0.9 and learning rate of 0.2. Grid search optimization produced an RMSE of 0.101 with a training cycle of 100, learning rate of 1, and momentum of 0.1. Evolutionary parameter optimization produced an RMSE of 0.103 with a learning rate of 0.029, momentum of 0.68, and training cycle of 86.
Based on the study, grid search parameter optimization produced a better RMSE than evolutionary parameter optimization. The activity addresses the stated goal of analysing JCI stock-price movement during COVID-19 through neural-network testing and parameter optimization. It provides an institutional research activity concerning the analysis of financial-market data during the pandemic.