An Enhanced Hidden Markov Model for Predicting Two Tier Web Page and Improving Accuracy

The weblog contains a substantial amount of information about user requests. To offer suggestions for websites, minimize latency, and engage in other online activities such as advertising, it is critical to be able to anticipate future requests made by users based on the pages they have seen in the past. These applications compromise on the intricacy of their models and the accuracy of their forecasts. A web page prediction model that builds and provides a classifier using sessions as training examples. The former system, which used 30 min as the default timeout, has been replaced with one that produces sessions by evaluating the average time spent on website access. Because sessions are utilized as training examples, this is done. Hidden Markov models have lately gained popularity for predicting the next page a user will visit on a website based on navigational history saved in a weblog. This usage-based technique has the potential for further 456improvement to improve forecast accuracy. The hidden Markov model with an autoregressive model is suggested as a way of improving web page prediction accuracy by 83% compared with the conventional hidden Markov model.