APPLYING DATA MINING TECHNIQUES TO INVESTIGATE ONLINE SHOPPER PURCHASE INTENTION BASED ON CLICKSTREAM DATA
Keywords:Clickstream data, data mining, machine learning, online shopping, purchase intention
Purpose: This study aims to analyse visitors' behaviour patterns to the online shop as a useful reference in e-commerce’s strategies.
Design/methodology/approach: The advanced analysis investigates online shoppers' intentions by analysing visitor behaviour patterns in the online shopping environment. Data mining techniques are applied to find useful information. This study applies feature selection to filter out unrepresentative features by analysing the visitor's clickstream data for one year and selecting representative features. This study examines several supervised machine learning algorithms to identify more accurate prediction performance.
Findings: This study finds the page value measured by Google Analytics directly indicates the most influential pages on online shops' visitors. Moreover, Neural Net was the fittest of accuracy and F score. Random Forest is the fittest in the ROC curve's visual as the robust prediction performances among other algorithms.
Originality: This paper explores feature selection through cross-validation for each classifier and compares the performance with accuracy, F score and the ROC curve as robust techniques to predict determinant online shopper purchase intention.
How to Cite
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.