Proxy caching is an effective method to improve response latency and reduce network traffic. There are two main alternative approaches that are automatically managed and tuned Web resources. Although both replacement and prefetching strategies very useful to Web caching but several replacement strategies remain slightly improvement. On the other hand, the challenge schemes focus on prefetching techniques. Prefetching becomes very popular by predict the possibility of which web object could be requested in near future. This paper present the prediction for Web caching model, which apply web mining and machine learning techniques for prediction in prefetching and caching. Interval period time of re-ordering Web object requested information from Web logs have been analyzed and apply statistical techniques to produce useful information to predict future requested. Our results show potential improvement to achieve better performance management.