通過網(wǎng)絡爬蟲獲取網(wǎng)絡空間數(shù)據(jù)以分析和預測物理世界的宏觀事件是近年來十分重要的研究方向。本論文面向旅游業(yè)需求研究了如何利用谷歌趨勢統(tǒng)計的網(wǎng)絡搜索信息來預測物理世界的真實游客數(shù)量。世界各地的旅游業(yè)在迅猛發(fā)展中:荷蘭的旅游業(yè)占國民生產(chǎn)總值的9%,其首都阿姆斯特丹是一個非常美麗的城市,有著郁金香、溝渠、游艇和各種展覽館。有許多著名的畫家都生活在阿姆斯特丹,游客可以在這里的畫廊里發(fā)現(xiàn)許多備受贊譽的杰作。阿姆斯特丹的旅游業(yè)對整個荷蘭的經(jīng)濟發(fā)展貢獻很大。因此,準確預測阿姆斯特丹的游客數(shù)量具有重要的實際意義。本文的研究思路是利用谷歌趨勢信息來預測阿姆斯特丹的旅游業(yè)需求。具體的,我們利用Touristjourney對搜索查詢詞進行拓展和篩選,然后通過GoogleSearch Query利用谷歌趨勢Google Trends返回的與查詢詞相關(guān)的搜索統(tǒng)計信息分析真實游客數(shù)量的相關(guān)性并訓練得到隱馬爾科夫模型;在測試階段,將搜索參數(shù)歸集到Google Trends中,獲得查詢詞列表和對應的Google Trends信息,進而通過訓練的隱馬爾科夫模型進行游客數(shù)量預測。這項研究發(fā)現(xiàn),谷歌趨勢提供的信息對于確定阿姆斯...

【文章頁數(shù)】：61 頁

【學位級別】：碩士

【文章目錄】：
ACKNOWLEDGEMENT
摘要
ABSTRACT
ABBREVIATIONS AND ACRONYMS
1. INTRODUCTION
    1.1. TOURIST JOURNEY
    1.2. PROBLEM STATEMENT
    1.3. Research Objectives
    1.4. RESEARCH QUESTIONS
    1.5 THESIS OUTLINES
2. LITERATURE REVIEW
    2.1. RELATED WORK
        2.1.1. Econometric Models
        2.1.2. Artificial Intelligence
        2.1.3. Artificial Neural Networks (ANN)
    2.2 CONTRIBUTION AND DIFFERENCE WITH PREVIOUS WORKS
    2.3. BRIEF OVERVIEW OF SEARCH ENGINE AND SOCIAL MEDIA
        2.3.1. Social Media
        2.3.2. Search Engine
    2.4. FORECASTING BY THE CITY OF AMSTERDAM
3. METHODOLOGY
    3.1. THE WORKHOW OF OUR METHOD
    3.2. DATA COLLECTION
    3.3. HIDDEN MARKOV MODEL
    3.4. ARTIFICIAL NEURAL NETWORK (ANN)
    3.5. VECTOR AUTO-REGRESSIVE (VAR)
    3.6. HIDDEN MARKOV MODEL AS A SOLUTION
        3.6.1. Keywords Extraction and Evaluation
        3.6.2. Hidden Markov Model Training
        3.6.3. Hidden Markov Model Prediction
4. EXPERIMENTS AND RESULTS
    4.1. EXPERIMENTAL SETUP
    4.2. STATISTICS OF EXTRACTED KEYWORDS
    4.3. RESULTS OF GRANGER CAUSALITY ANALYSIS
    4.4. PREDICTION PERFORMANCE COMPARISON
5. CONCLUSION AND DISCUSSION
    5.1. CONCLUSIONS
    5.2. DISCUSSION
    5.3. LIMITATIONS AND FUTURE RESEARCH
REFERENCES
AUTHOR PROFILE
DATASET FOR THE MASTER'S THESIS



本文編號：3735074