Abstract
Weather observation data produced by the Meteorology, Climatology and Geophysics Agency (BMKG) has multivariate characteristics, so it requires an analysis method that is able to identify patterns based on similarities in data characteristics. This research aims to analyze and visualize daily and monthly weather parameter patterns in the East Jakarta area using the K-Means Clustering algorithm. The dataset used is daily observation data from the Halim Perdanakusuma Meteorological Station for the period January–December 2025, totaling 365 observations with six meteorological parameters, namely TN, TX, TAVG, RH_AVG, RR, and FF_AVG. The research stages include merging data, handling missing values using median imputation, data normalization using StandardScaler, data exploration, determining the optimal number of clusters using the Elbow Method and Silhouette Score, followed by the clustering process using K-Means. The evaluation results show that there are three optimal clusters (K=3), which represent variations in the daily weather characteristics of East Jakarta. The clustering results identified different characteristics, namely Cluster 0 (Hot Dry Weather), Cluster 1 (Cool Cloudy Weather), and Cluster 2 (Heavy Rainy Weather). Visualization in the form of correlation heatmaps, time series graphs, scatter plots, box plots and bar charts provide informative representations of weather parameter distribution patterns, making it easier to interpret analysis results. This research proves that K-Means Clustering is used as an effective exploratory method for identifying weather parameter patterns and visualization of meteorological analysis.
