Web Traffic Time Series Forecasting For Wikipedia Pages

Abstract

Web traffic time series forecasting for Wikipedia pages is a critical task for content creators, administrators, and analysts. Predicting the future page views of Wikipedia articles enables better resource allocation, content optimization, and decision-making. In this paper, we explore various techniques and models for forecasting Wikipedia page views, considering the unique challenges posed by the diversity of topics and the dynamic nature of online content consumption. We present a comprehensive analysis of the data preprocessing, feature engineering, and modeling steps involved in accurate time series forecasting. The performance of different forecasting models is evaluated using real-world Wikipedia page view data, and we discuss the practical implications of our findings. This study adds to the more extensive field of web traffic analysis and provides insights into the predictability of online user behavior for diverse content domains.