ABSTRACT
This paper presents an approach of dimension reduction using Principal Component Analysis (PCA) to the LQ 45 index of the Indonesia Stock Exchange. Specifically, this paper has several objectives. First, this paper aims to reduce the complexity of the data into a small number of principal components, which forms the linear combinations of prices that minimizes information loss and can explain as much as possible the variation in price data. Second, this paper aims to find the relative importance of the variables in the component. Third, this paper also aims to find whether certain stocks group with others and whether there are some common attributes between certain stocks for us to further study to get more insights. This paper used 44 of 45 stocks included in the LQ 45 index. The first finding is that there are seven principal components formed out of the 45 stocks. For the first principal component, the five largest factor loadings are PTPP, BRPT, WSKT, PGAS, and BBCA, respectively. For the second principal component, the five largest factor loadings are EMTK, ANTM, TBIG, HRUM, and TINS, respectively. Third, the results show that some stocks tend to cluster with others. For the clustering analysis, we can conclude that the PCA can show that some stocks do tend to cluster together and we are able to identify three clusters: cluster of stocks with loading close to zero, meaning they do not contribute much for the variation of the LQ 45 index. Clusters with factor loadings close to ±0.5 show moderate contribution to the variation of LQ 45 index. The third cluster tends to have larger factor loadings, either negative or positive.
