Clustering is a method to group data into clusters based on their natural structure. A new algorithm called Grid-K-means was created to improve clustering efficiency and accuracy for big data. This algorithm uses grids to determine cluster centers and evaluate clustering quality. Testing on different datasets showed that Grid-K-means is faster and more accurate than traditional methods. The new clustering validity index (BCVI) introduced in this study is better at determining the optimal number of clusters for large datasets compared to existing indexes.