Climate change occurs in the atmosphere over a long period due to the influence of the sun, oceans, clouds, ice, land, and living organisms on each other. This research used the Principal Component Analysis (PCA) model compounded with Vector Autoregressive Integrated (VARI) called the PCA-VARI model to determine climate change. PCA reduces correlated climate data to uncorrelated data expressed as main components containing a linear combination of initial variables. In the time series model, a non-stationary multivariate comprises more than two variables that influence each other, using differencing processes. A variety of two models was used simultaneously to forecasting future climate data. Analysis of climate parameters uses ten measurements variable located in five areas, namely Lembang, Bogor, Tasikmalaya, Sukabumi, and Indramayu, for twenty years, using POWER NASA Agro-climatology datasets. The methodology follows the Knowledge Discovery in Databases (KDD) in data mining for integrated PCA with VARI and post-processing using visualization by Impulse Response Function (IRF). The result of forecasting in the PCA-VARI model using IRF in the next six months showed that the effect of location climate on the response of other regions with changes in standard deviation is similar to adjacent locations. Meanwhile, the responses obtained varied based on the observation time for the five areas that are not close.