Towards Efficient Forecasting of Web Articles using Fibonacci Median

As the number of internet users is increasing day by day, the traffic over the internet is getting q uadrupled in its volume. This work tries to forecast the traffic o ver t he internet by presenting an approach based on the log transformation and the usage of the Fibonacci median for the feature selection process. The results of the technique have been compared with the standard LSTM (Long Short Term Memory) and ARIMA (Auto-Regressive Integrated Moving Average) models. It has been observed from the analysis that our proposed approach provides better results compared to the LSTM technique in terms of lower RMSE (Root Mean Square Error) and SMAPE (Symmetric Mean Absolute Percentage Error) values. However, these resulting errors are still more than that of ARIMA model. This technique can further enhance the performance of already existing methods and open prospects for a better user experience on the internet.