Pre-emptive Caching of Video Content Using Predictive Analysis

Pre-emptive caching is a technique to pre-fetch data based on the outcome of algorithmic predictions. In this paper, we use machine learning models that account for time-based trends instead of metadata to estimate a popularity score integrated with existing caching schemes to pre-fetch trending videos or replace cached content with newer videos having higher in-demand user requests. The paper primarily focuses on regression models for prediction as they are faster to train which is crucial for a resource-intensive task like caching.