Wikipedia Web Traffic Prediction

Comparative time-series forecasting study on 145,000 Wikipedia articles — co-authored research paper under revision

Research Paper (Under Revision) Co-authored with Prof. Sandhya Avasthi, ABES Engineering College, India 2025

Overview

Conducted a comparative study of 7 time-series forecasting models on a 145,000-article Wikipedia Kaggle dataset, evaluated using RMSE as the primary performance metric.

Models Compared

  • Moving Average
  • Exponential Smoothing
  • ARMA
  • ARIMA
  • Auto-ARIMA
  • Prophet
  • LSTM

Key Finding

Identified Moving Average as the best-performing model — demonstrating that simpler statistical models outperformed deep learning (LSTM) on this task, providing practical model selection insights for large-scale web traffic forecasting.

Collaboration

Co-authored with international academic collaborator across two institutions (BatStateU, Philippines & ABES Engineering College, India). Submitted to journal, currently under revision.

Tech Stack: R · Time-Series Analysis · ARIMA · Prophet · LSTM · Kaggle