Predict Tourism Demands Using Google Trends

Predict macro events by using social media analysis and search engines is an interesting topic in recent years;the aim of this research demonstrates how information from Google Trends can be utilized to predict tourism demands.When a tourist interacts with the internet through a search engine,a website,or a social media platform,the traces of the interaction can be captured,stored,and analyzed.The tourism sector is still growing in whole the world;in the Netherlands,the tourism industry takes 9%of the Dutch domestic product.The reason why we selected Amsterdam as a case study is that research on most visited cities on the world showed how Amsterdam become a standout among the most visited city populations on the planet,all things considered,Amsterdam is a beautiful city it induces pictures of tulips,trenches,cycling courses,and museums.The city is riches made it a primary for attraction a chance to encounter better things throughout everyday life.Numerous well-known painters lived in Amsterdam,and you can discover much-admired work of art.these days Amsterdam travel industry depends on its picture as a gathering capital.Amsterdam tourism industry generally contributes to its riches today.Related works demonstrated that Google Trends have some values to predict the tourism industry,our new idea for this study is to utilize our model with Google Trends data to predict the tourism demands in Amsterdam and to compare our method Hidden Markov Model(HMM)with existing methods.The tourist journey was utilized to subtract search query terms deductively,and the other way around tourist statistics connected to Google Trends with the utilization of Google search query(keywords)to which search parameters are assigned,a list of search query related to travel and the tourism industry utilized.This research found that information given by Google Trends can be valuable for determining tourists number in Amsterdam if the appropriate keywords utilized,the proposed HMM model was applied to predict visitors using historical data from the electronic database Central Bureau Statistic(CBS)StatLine 2018 and Google Trends.The search engine user needs to utilize search queries(keywords)identified with the tourism industry in Amsterdam will be extracted using application programming interface(API),to apply this procedure we proposed Hidden Markov Model demonstrate as an answer for predict next month tourists number arrival in Amsterdam.The following month tourists number arrival in Amsterdam is taken as the objective prediction,related with the three info highlights which are the date,tourists number,and region from Google Trends informational index,the output is the number of tourists in Amsterdam.From May 29th,2016,up to 31st December 2018 aggregate of 265,307 search engine users from Five nations(United Kingdom,Germany,France,Belgium,Sweden),utilized with six predictive power search queries(Amsterdam,Hotel Amsterdam,visit Amsterdam,travel Amsterdam,city trip Amsterdam,Holiday Amsterdam).We have trained and tested data by the Hidden Markov Model to predict tourists number in Amsterdam,and compare our method with other existing methods.Comparison with existing methods using Google Trends as HMM performance is adjustable by the coverage bound;we set 3 values between 1.0 and 0.1 for to evaluate HMM,a smaller CB value means that HMM puts more control on prediction output,which leads to a lower error rate.For the two existing methods which are Vector Autoregressive(VAR)and Artificial Neural Networks(ANN),we tune their parameters to get the best results.The advantage of using query trends to predict tourist data in real time,first query trends can use updated information up to the day before the forecast computation,which could potentially be highly valuable in this context due to the lags in the publication of the official statistics.Our experiments over real data from CBS StatLine demonstrate that our method not only outperforms the traditional and existing methods but also provides controllability to tourism prediction.