Clickstream Data Research Articles

Machine-Learning-Based Approach for Anonymous Online Customer Purchase Intentions Using Clickstream Data

Since online shopping has become an important way for consumers to make purchases, consumers have signed up to e-commerce platforms to shop online. However, retailers are beginning to realise the critical role of predicting anonymous consumer purchase intent to improve purchase conversion rates and store profitability. Therefore, this study aims to investigate the prediction of anonymous consumer purchase intent. This research presents a machine learning model (MBT-POP) for predicting customer purchase behaviour based on multi-behavioural trendiness (MBT) and product popularity (POP) using 33,339,730 clicks generated from 445,336 sessions of real e-commerce customers. The results show that the MBT-POP model can effectively predict the purchase behaviour of anonymous customers (F1 = 0.9031), and it achieves the best prediction result with a sliding window of 2 days. Compared to existing studies, the MBT-POP model not only improves the model performance, but also compresses the number of days required for accurate prediction. The present research has argued that product trendiness and popularity can significantly improve the predictive performance of the customer purchase behaviour model and can play an important role in predicting the purchase behaviour of anonymous customers.

DATA AND WEB MINING

Data and web mining involves the use of data mining techniques to extract information from the web in order to gain insights into online behavior, customer preferences, and trends. The process requires data preprocessing to clean and prepare the data for analysis, and pattern recognition to identify relationships and patterns within the data. The benefits of data and web mining include improved decision-making and business growth, while challenges include the large volume of data and ensuring data quality. Web mining can be divided into three main categories: web content mining, web structure mining, and web usage mining. Web content mining involves extracting information from the content of web pages, such as text, images, and videos. Web structure mining involves analyzing the links between web pages to identify patterns and relationships. Web usage mining involves analyzing user behavior on the web, such as clickstream data and session logs, to extract useful information. KEYWORDS: data mining, web mining, information extraction, data preprocessing, pattern recognition, online behavior, customer preferences,

Maximizing Online Revisiting and Purchasing: A Clickstream-Based Approach to Enhancing Customer Lifetime Value

ABSTRACT Online retailers are increasingly focused on maintaining a long-term relationship with customers, encouraging repeat visits rather than single-time purchases to increase customer lifetime value. To help retailers maximize the probabilities of customers’ revisiting and purchasing, we develop a two-stage model to better characterize and predict these two fundamental customer activities. In the first stage, we characterize the propensity of a customer revisiting the retailer’s website. In the second stage, we develop a stochastic model that predicts revisits while also incorporating individual customer heterogeneity in exerted search effort during repeated visits. This heterogeneity is based on individual customer preferences in the choice of consideration sets, product information, pricing, and the search environment. Using customer level clickstream data, we show that our approach is not only better at predicting repeat customer visits, compared to existing methods, but also explainable and managerially interpretable. Most importantly, using computationally efficient simulation-based prescriptive analytics, we leverage our modeling approach to propose practical intervention strategies that maximize the joint likelihoods of customers revisiting and purchasing at the individual customer level.

Shopping Hard or Hardly Shopping: Revealing Consumer Segments Using Clickstream Data

The recent rise of big data analytics is transforming the apparel retailing industry. E-retailers, for example, effectively use large volumes of data generated as a result of their day-to-day business operations data to aid operations and supply chain management. Although logs of how consumers navigate through an e-commerce website are readily available in a form of clickstream data, clickstream analysis is rarely used to derive insights that can support marketing decisions, leaving it an under-researched area of study. Adding to this research stream by exploring the case of a U.K.-based fast-fashion retailer, this article reveals unique consumer segments and links them to the revenue they are capable of generating. Applying the partitioning around medoids algorithm to three random samples of 10 000 unique consumer visits to the e-commerce site of a fast-fashion retailer, six consumer segments are identified. This article shows that although the “mobile window shoppers” segment consists of the largest consumer segment, it attracts the lowest revenue. In contrast, “visitors with a purpose,” although one of the smallest segments, generates the highest revenue. The findings of this article contribute to marketing research and inform practice, which can use these insights to target customer segments in a more tailored fashion.

Measuring Benefits from New Products in Markets with Information Frictions

I study how much consumers benefit from new products in markets with information frictions. I analyze new products in the U.S. hard drive market, which is characterized by ample product innovation. Using unique click-stream data, I measure the magnitude of two frictions, category consideration and costly search, and show that both play a crucial role in shaping consumer demand. To estimate consumer surplus from new products, I develop a search model that captures both frictions and propose a novel Bayesian estimation method to recover its parameters. I then show that ignoring information frictions leads researchers to underestimate the consumer surplus from new hard drives because it appears that consumers do not value the combinations of attributes these hard drives offer. Partly eliminating frictions, through marketing efforts or market-wide transparency initiatives, can help consumers to more fully internalize the benefits of new product launches. This paper was accepted by Matthew Shum, marketing. Supplemental Material: The data files and online appendices are available at https://doi.org/10.1287/mnsc.2023.4729 .

Model selection uncertainty and stability in beta regression models: A study of bootstrap-based model averaging with an empirical application to clickstream data

Statistical model development is a central feature of many scientific investigations with a vast methodological landscape. However, uncertainty in the model development process has received less attention and is frequently resolved nonrigorously through beliefs about generalizability, practical usefulness, and computational ease. This is particularly problematic in settings of abundant data, such as clickstream data, as model selection routinely admits multiple models and imposes a source of uncertainty, unacknowledged and unknown by many, on all postselection conclusions. Regression models, based on the beta distribution, are a class of nonlinear models, attractive because of their great flexibility and potential explanatory power, but have not been investigated from the standpoint of multimodel uncertainty and model averaging. For this reason a formalized tool that can combine model selection uncertainty and beta regression modeling is presented in this work. The tool combines bootstrap model averaging, model selection, and asymptotic theory to yield a procedure that can perform joint modeling of the mean and precision parameters, capture sources of variability in the data, and achieve more accurate claims of estimate precision, variable importance, generalization performance, and model stability. Practical utility of the tool is demonstrated through a study of model selection consistency and variable importance in average exit and bounce rate statistical models. This work emphasizes the necessity of a departure from the all-too-common practice of ignoring model selection uncertainty and introduces an accessible technique to handle frequently neglected aspects of the modeling pipeline.

Advancing the marketing‐operations interface in omnichannel retail

Advancing the marketing‐operations interface in omnichannel retail

Fewer Clicks, More Purchases

E-commerce retailers are increasingly faced with challenges of finding ways to provide a seamless shopping experience to customers. The checkout process and its related touch points are especially critical in shaping the customer experience. We study the impact of adopting one-click buying, a feature that reduces the number of steps required to place a purchase order to a single click, on subsequent customer behavior. Using quasi-experimental data over a period of 35 months from an online retailer before and after the launch of one-click buying, we find adopting one-click buying is effective in lifting customer purchases and does so by making treated customers purchase more often and more items. The impact of adopting one-click buying on customer purchases after adoption is economically significant, persistent over time, and heterogeneous across customers. Analyzing clickstream data of customer activity online and purchases across product categories, we provide evidence that the increase in purchases is driven by richer engagement through both more visits to the website and more page views on visit and the expansion of purchases across categories. We discuss the implications of our findings for customer experience and targeting. This paper was accepted by Duncan Simester, marketing. Supplemental Material: The data files are available at https://doi.org/10.1287/mnsc.2023.4716 .

The Effect of Self-Regulated Learning in Online Professional Training

With the rapid expansion of mobile, blended, and seamless learning, researchers claim two factors, lack of self-discipline and poor time management, adversely impact learning performance. In online educational environments, reduced social interactions and low engagement levels generate high dropout rates. Self-regulated learning (SRL), the individual ability to check progress toward a goal and manage learning behavior, appears critical to adult online learning success. Clickstream data can observe, record, and evaluate patterns of users' real-time learning behavior in an online learning environment. Linking clickstream data with performance outcomes allows researchers to assess online learning behaviors and academic performance. The guiding research question was: Are students who apply SLR strategies more likely to demonstrate mastery of knowledge and skills in a self-directed e-learning context? Clickstream data and performance measures were analyzed to explore whether task and cognitive conditions influence how SLR strategies are applied in online training.

Learning analytics on video-viewing engagement in a flipped statistics course: Relating external video-viewing patterns to internal motivational dynamics and performance

This study attempts to advance the comprehension of clickstream data in video-based learning environments and interpret learning motivation from these large-volume and unstructured data. Three hundred and fifty video-learning records from 47,044 video viewing operations were obtained from 47 graduate students. Four learning improvement profiles were derived from a statistics course's pre- and post-test scores: the Advanced, the Diligent, the Indifferent, and the Persistent. The results indicated that the Diligent and the Persistent paused videos frequently. These students reflected the need with these self-paced breaks to take notes according to the retrospective interview. The Advanced demonstrated the highest SkippingBackward frequency with a small SkippingBackward time ratio, revealing confidence in searching the desired clips. In contrast, the Indifferent exhibited the least frequency but the largest time ratio for SkippingBackward, implying distraction problems. In addition, this study showed that the Pause time ratio was indirectly related to weekly quiz scores via autonomous motivation with marginal significance. Based on these results, we demonstrate that learning motivation can be revealed from dynamic clickstreams in video-based learning (i.e., the interaction between students and learning contexts), supporting its dynamic developmental mechanism. The authors suggest that instructors offer on-demand instructional materials and implement top-down video-viewing strategies or digital prompts to support and encourage autonomy in video-based learning.

The version effect of apps and operating systems in mobile commerce

Different versions of mobile operating systems or shopping apps enable different functionalities and information flows, thus creating various mobile shopping environments. In general, up‐to‐date versions provide better information flows and richer functionalities. However, mobile operating systems and shopping apps affect consumer behavior through different mechanisms. An up‐to‐date mobile operating system reduces the system response time, while an up‐to‐date shopping app improves algorithms for better search accuracy. The former encourages consumer explore and search, whereas the latter improves consumer search efficiency and reduces chances for consumers to discover more products. Using a unique large‐scale clickstream data set from a mobile commerce retailer, we examine the effect of mobile operating system and app versions on consumer search and impulse purchase behaviors in mobile commerce. Our results show that consumers with an up‐to‐date mobile operating system or a previous version of a shopping app conduct more searches in terms of increased product page views and time spent on product pages, which results in a higher probability of consumer impulse purchase. However, consumers' search for individual products is not affected by versions. Surprisingly, though more page views or time spent may boost purchases, we find that consumer search affects impulse purchases nonlinearly in that page views and time spent have a decreasing rate of impact. Our computations show that using an up‐to‐date operating system increases consumer search activities by 60.45 product pages or 521.30 seconds spent browsing compared to using a previous version of an operation system. Using an up‐to‐date app decreases consumer search activities by 129.34 product pages or 1446.48 seconds spent browsing compared to using a previous version of a mobile app.

Clicking through the Clickstream: A Novel Statistical Modeling Approach to Improve Information Usage of Clickstream Data by E-Commerce Entities

Success or failure of an E-commerce platform is often reduced to its ability to maximize the conversion rate of its visitors. This is commonly regarded as the capacity to induce a purchase from a visitor. Visitors possess individual characteristics, histories, and objectives which complicate the choice of what platform features that maximize the conversion rate. Modern web technology has made clickstream data accessible allowing a complete record of a visitor’s actions on a website to be analyzed. What remains poorly constrained is what parts of the clickstream data are meaningful information and what parts are accidental for the problem of platform design. In this research, clickstream data from an online retailer was examined to demonstrate how statistical modeling can improve clickstream information usage. A conceptual model was developed that conjectured relationships between visitor and platform variables, visitors’ platform exit rate, boune rate, and decision to purchase. Several hypotheses on the nature of the clickstream relationships were posited and tested with the models. A discrete choice logit model showed that the content of a website, the history of website use, and the exit rate of pages visited had marginal effects on derived utility for the visitor. Exit rate and bounce rate were modeled as beta distributed random variables. It was found that exit rate and its variability for pages visited were associated with site content, site quality, prior visitor history on the site, and technological preferences of the visitor. Bounce rate was also found to be influenced by the same factors but was in a direction opposite to the registered hypotheses. Most findings supported that clickstream data is amenable to statistical modeling with interpretable and comprehensible models.

Improving Shopping Mall Revenue by Real-Time Customized Digital Coupon Issuance

With the development of big data and deep learning technology, big data and deep learning technology have also been applied to the marketing field, which was a part of business administration. Customer churn management is one of the most important areas of marketing. In this paper, we proposed a method to prevent customer churn and increase purchase conversion rate by issuing customized discount coupons to customers with high churn rate based on big data in real time. After segmenting customer segments with two-dimensional segment analysis, a real-time churn rate estimation model based on clickstream data was generated for each segment. After that, we issued customized coupons to our customers. Finally, we tested the conversion rate and sales growth. A two-dimensional cluster analysis-based churn rate estimation combined with a recommendation system was found to be significantly more useful than the respective simple models. Using this proposed model, it is possible to increase sales by automatically estimating the customer’s churn probability and shopping propensity without the burden of marketing costs in the online shopping mall.

Predicting Student Performance Using Clickstream Data and Machine Learning

Student performance predictive analysis has played a vital role in education in recent years. It allows for the understanding students’ learning behaviours, the identification of at-risk students, and the development of insights into teaching and learning improvement. Recently, many researchers have used data collected from Learning Management Systems to predict student performance. This study investigates the potential of clickstream data for this purpose. A total of 5341 sample students and their click behaviour data from the OULAD (Open University Learning Analytics Dataset) are used. The raw clickstream data are transformed, integrating the time and activity dimensions of students’ click actions. Two feature sets are extracted, indicating the number of clicks on 12 learning sites based on weekly and monthly time intervals. For both feature sets, the experiments are performed to compare deep learning algorithms (including LSTM and 1D-CNN) with traditional machine learning approaches. It is found that the LSTM algorithm outperformed other approaches on a range of evaluation metrics, with up to 90.25% accuracy. Four out of twelve learning sites (content, subpage, homepage, quiz) are identified as critical in influencing student performance in the course. The insights from these critical learning sites can inform the design of future courses and teaching interventions to support at-risk students.

A hybrid recommender system based on customer behavior and transaction data using generalized sequential pattern algorithm

In the future, the quality of product suggestions in online retailers will influence client purchasing decisions. Unqualified product suggestions can result in two sorts of errors: false negatives and false positives. Customers may not return to the online store as a result of this. By merging sales transaction data and consumer behavior data in clickstream data format, this work offers a hybrid recommender system in an online store utilizing sequential pattern mining (SPM). Based on the clickstream data components, the product data whose status is only observed by consumers is assessed using the simple additive weighting (SAW) approach. Products with the two highest-ranking values are then coupled with product data that has been purchased and examined in the SPM using the generalized sequential pattern (GSP) method. The GSP algorithm produces rules in a sequence pattern, which are then utilized to construct product suggestions. According to the test results, product suggestions derived from a mix of sales transaction data and consumer behavior data outperform product recommendations generated just from sales transaction data. Precision, recall, and F-measure metrics values rose by 185.46, 170.83, and 178.43%, respectively.

Barriers and facilitators to use of a digital clinical decision support tool: a cohort study combining clickstream and survey data

ObjectivesThis research aimed to understand the barriers and facilitators clinicians face in using a digital clinical decision support tool—UpToDate—around the globe.DesignWe used a mixed-methods cohort study design that enrolled 1681...

The use of video clickstream data to predict university students' test performance: A comprehensive educational data mining approach.

Video clickstream behaviors such as pause, forward, and backward offer great potential for educational data mining and learning analytics since students exhibit a significant amount of these behaviors in online courses. The purpose of this study is to investigate the predictive relationship between video clickstream behaviors and students' test performance with two consecutive experiments. The first experiment was performed as an exploratory study with 22 university students using a single test performance measure and basic statistical techniques. The second experiment was performed as a conclusive study with 16 students using repeated measures and comprehensive data mining techniques. The findings show that a positive correlation exists between the total number of clicks and students' test performance. Those students who performed a high number of clicks, slow backward speed or doing backwards or pauses achieved better test performance than those who performed a lower number of clicks, or who used fast-backward or fast-forward. In addition, students' test performance could be predicted using video clickstream data with a good level of accuracy (Root Mean Squared Error Percentage (%RMSE) ranged between 15 and 20). Furthermore, the mean of backward speed, number of pauses, and number/percentage of backwards were found to be the most important indicators in predicting students' test performance. These findings may help educators or researchers identify students who are at risk of failure. Finally, the study provides design suggestions based on the findings for the preparation of video-based lectures.

How Appreciation and Attention Affect Contributions to Electronic Networks of Practice

ABSTRACT We conducted three studies to examine how two types of user-generated feedback, appreciation and attention, affect users’ decisions to make voluntary knowledge contributions to electronic networks of practice (ENPs). Appreciation is reflected in positive ratings, votes, and helpfulness evaluations. Attention is reflected in the number of views of contributed content. The first study used clickstream data from a college application ENP in China, where information seekers can read posted information asynchronously and request synchronous consultations with volunteers. The second study was a controlled online experiment in the United States where we assessed users’ willingness to answer questions in a college application ENP asynchronously. The third study examined knowledge contribution across a diverse set of topics using a well-established ENP that serves more than 100 countries. In all three studies, the results consistently show that greater appreciation increased continued knowledge contribution, but greater attention without sufficient appreciation negatively affected contributions. Our findings show the theoretically intertwined nature of attention and appreciation and offer insights for the design and management of ENP feedback systems to encourage user contributions.

Research on intelligent information system of user intelligent behavior data based on computer big data

This paper collects and analyzes users' online behaviors through computer big data technology, uses computer big data intelligent analysis system to analyze users' online shopping behavior, and uses information-based data analysis system to detect consumers' online shopping needs under the e-commerce platform. The main technique used in this paper is the computer browser log mining method. In the user's click stream data, the function keys of Tmall and Taobao webpages are used as data information for classification and collection. This paper uses the Bisecting K-means clustering algorithm to mine the state of interest. Finally, the feature maps of interests and behaviors are summarized. By processing four typical types of e-commerce user demand status, including background management type, continuous search type, product browsing type and information search data, and clustering based on page type, an effective method for dynamically changing demand judgment is obtained. The state of online shoppers is analyzed through data processing, which also proves the effectiveness of the computer intelligent information system.

The effects of assessment design on academic dishonesty, learner engagement, and certification rates in MOOCs

AbstractBackgroundMassive Open Online Courses (MOOCs) have touted the idea of democratizing education, but soon enough, this utopian idea collided with the reality of finding sustainable business models. In addition, the promise of harnessing interactive and social web technologies to promote meaningful learning was only partially successful. And finally, studies demonstrated that many learners exploit the anonymity and feedback to earn certificates unethically. Thus, establishing MOOC pedagogical models that balance open access, meaningful learning, and trustworthy assessment remains a challenge that is crucial for the field to achieve its goals.ObjectivesThis study analysed the influence of an MOOC assessment model, denoted the Competency Exam (CE), on learner engagement, the level of cheating, and certification rates. At its core, this model separates learning from for‐credit assessment, and it was introduced by the MITx Biology course team in 2016.MethodsWe applied a learning analytics methodology to the clickstream data of the verified learners (N = 559) from four consecutive runs of an Introductory Biology MOOC offered through edX. The analysis used novel algorithms for measuring the level of cheating and learner engagement, which were developed in the previous studies.Results and ConclusionsOn the positive side, the CE model reduced cheating and did not reduce learner engagement with the main learning materials – videos and formative assessment items. On the negative side, it led to procrastination, and certification rates were lower.ImplicationsFirst, the results shed light on the fundamental connection between incentive design and learner behaviour. Second, the CE provides MOOC designers with an ‘analytically verified’ model to reduce cheating without compromising on open access. Third, our methodology provides a novel means for measuring cheating and learner engagement in MOOCs.