Prediction of Indonesia composite index using BI rate with least square spline nonparametric regression

Predicting the Indonesia Composite Index (ICI) is essential due to its crucial role in the Indonesian economy. Its fluctuations significantly affect investor confidence in the country’s economic prospects. Numerous factors can influence ICI movements, including the BI rate variations, which affect companies’ funding and investment returns. Time series data, such as ICI data, consist of observations gathered at specific intervals. The data typically show a fluctuating pattern, making it challenging to predict using a parametric approach. In such cases, a nonparametric approach, known for its flexibility because it does not require certain assumptions like parametric methods, is applicable. This research utilizes a nonparametric regression approach using a Least Square Spline Estimator (LSSE) to predict ICI based on the BI Rate. The spline method was chosen for its flexibility in modeling by placing knot points that allow the model to fit the data accurately. The selection of optimal knot points in this study was based on the smallest Generalized Cross Validation (GCV) value. This study’s optimal knot points were chosen based on the smallest Generalized Cross Validation (GCV) value. The goodness of fit in this study was carried out by calculating the Mean Absolute Percentage Error (MAPE) value. Based on the training data, the MAPE value obtained was 7.03 and 1.75 was the MAPE value for the testing data. So, the overall MAPE value is 4.39 or less than 10, which indicates that the resulting model is highly predictable. Based on the BI Rate, the model developed in this research can predict ICI.