Clickstream Data Analysis Research Articles

A clickstream data analysis of Chinese academic library OPAC users' information behavior

Chinese academic libraries have been devoting great effort into introducing next-generation online public access catalogs (OPACs) in order to provide a better user experience. However, there is a lack of empirical research on their usage. In this study, a transaction log file from a typical next-generation OPAC, the Wuhan University Library OPAC, formed the basis for an investigation of users' information behavior, using a clickstream data analysis framework. The 26,732,368 clickstream records in the original log file were cleaned, parsed, coded, then analyzed at the footprint, movement, and pathway levels. The results showed that the users relied heavily on the single-box simple search interface, seldom involved themselves in an exploratory search process, and preferred page navigation over search refinement when interacting with search results. The OPAC was used as a lookup tool to locate known academic resources and as a personal information management tool to utilize basic library services, instead of as a discovery tool as intended.

Serious Game Leverages Productive Negativity to Facilitate Conceptual Change in Undergraduate Molecular Biology

We designed a serious game, MolWorlds, to facilitate conceptual change about molecular emergence by using game mechanics (resource management, immersed 3rd person character, sequential level progression, and 3-star scoring system) to encourage cycles of productive negativity. We tested the value-added effect of game design by comparing and correlating pre- and post-test misconceptions, interaction statistics, and engagement in the game with an interactive simulation that used the same graphics and simulation system but lacked gaming elements. We tested first-, second-, and third-year biology students' misconceptions at the beginning and end of the semester (n = 526), a subset of whom played either the game (n = 20) or control (n = 20) for 30 minutes prior to the post-test. A 3x3 mixed model ANOVA revealed that, while educational level (first-, second-, or third-year biology) did not influence misconceptions from pre-test to post-test, the intervention type (no intervention, simulation, or game) did (p<.001). Pairwise comparisons showed that participants exposed to the interactive simulation (p = .007), as well as those exposed to the game (p<.001), lost significantly more misconceptions in comparison to those who did not receive any intervention, while adjusting for educational level. A trending difference was found between the simulation group and the gaming group (p = .084), with the gaming group resolving more misconceptions. Quantitative analysis of click-stream data revealed the greater exploratory freedom of the control simulation, with greater accessibility to individuals who do not play games on a regular basis. However, qualitative analysis of gameplay data showed that MolWorlds-players experienced significantly more instances of productive negativity than control-users (p<.001) and that a trending relationship exists between the quality of productively negative events and lower post-test misconceptions (p = .066).

Distributed coordinate descent for generalized linear models with regularization

Generalized linear model with L 1 and L 2 regularization is a widely used technique for solving classification, class probability estimation and regression problems. With the numbers of both features and examples growing rapidly in the fields like text mining and clickstream data analysis parallelization and the use of cluster architectures becomes important. We present a novel algorithm for fitting regularized generalized linear models in the distributed environment. The algorithm splits data between nodes by features, uses coordinate descent on each node and line search to merge results globally. Convergence proof is provided. A modifications of the algorithm addresses slow node problem. For an important particular case of logistic regression we empirically compare our program with several state-of-the art approaches that rely on different algorithmic and data spitting methods. Experiments demonstrate that our approach is scalable and superior when training on large and sparse datasets.

Using Sessions from Clickstream Data Analysis to Uncover Different Types of Twitter Behaviour

While much is known about how Twitter is used for specific tasks or by particular groups of users, we understand surprisingly little about how the service is used generally on a daily basis. To learn more about general Twitter behaviour we perform a cluster analysis on a rich set of longitudinal interaction log data describing interactions 44 users had with the Twitter website over a 5 month period. We report on and interpret 5 clusters representing common usage patterns with the service.

Website Clickstream Data Visualization Using Improved Markov Chain Modelling In Apache Flume

Clickstream data analysis is considered as the process of collecting, analysing and reporting the aggregate data about the web pages a visitor clicks. Visualizing the clickstream data has gained significant importance in many applications like web marketing, customer prediction, product management, etc. Most existing works employ different tools for visualizing along with techniques like Markov chain modelling. However the accuracy of the methods can be improved when the shortcomings are resolved. Markov chain modelling has problems of occlusion and unable to provide clear display of data visualizing. These issues can be resolved by improving the Markov chain model by introducing a heuristic method of Kolmogorov– Smirnov distance and maximum likelihood estimator for visualizing. These concepts are employed between the underlying distribution states to minimize the Markov distribution. The proposed model named as WebClickviz is performed in Hadoop Apache Flume which is a highly advanced tool. The clickstream data visualization accuracy can be improved when Apache Flume tools are used. The performance evaluation are made on a specific website clickstream data which shows the proposed model of visualization has better performance than existing models like VizClick.

Using Learning Analytics to Characterize Student Experimentation Strategies in the Context of Engineering Design

Engineering design is a complex process. The design process cannot be assessed based solely on a product or as a simple test because there is no single perfect design for a problem. An important design strategy is the conduction of experiments. Informed designers carry out experiments and use their outcomes to inform their next steps. On the other hand, beginning designers do little or no experiments, and the few experiments they do involve confounding variables. These behaviours that differentiate beginning and informed designers are not easy to assess in educational settings because they occur throughout the design process. This paper proposes and evaluates a model to analyze student interactions with a CAD tool in order to identify and characterize the different strategies students use to conduct experiments. A two-fold study is carried out to validate the model. The first phase uses the clickstream data of 51 middle school students working on a design project to create a net-zero energy house. The analysis of clickstream data is compared to a qualitative analysis of an open-ended posttest. The second phase correlates the number of experiments students did to the student prototype quality. The results suggest that the proposed model can be used to identify, characterize, and assess student strategies to conduct experiments.

인구통계특성 기반 디지털 마케팅을 위한 클릭스트림 빅데이터 마이닝

인구통계특성 기반 디지털 마케팅을 위한 클릭스트림 빅데이터 마이닝

Using clickstream data to enhance business‐to‐business web site performance

PurposeThis paper is designed to illustrate how clickstream data, collected from a B2B web site and then analyzed using web analytics software, can be used to evaluate and improve B2B web site performance. A number of issues in the application of clickstream data and web analytics software are to be identified and discussed.Design/methodology/approachA case study approach is used to present some of the technical issues in the field of web analytics and to demonstrate their value in B2B web site management. Three field experiments, focusing on incorporating ways to discourage shopping‐cart abandonment and the use of two different free‐shipping promotions, were used as the basic research method for collecting the data. Web traffic conversion funnels are used to conduct the analysis and present the findings.FindingsThe analysis of clickstream data using web analytics procedures serves as a useful tool in the enhancement of a B2B web site by investigating how visitors move through the web site conversion process and complete their purchase. Improved sales result from each of the three field experiments.Research limitations/implicationsResearchers may use the paper as evidence that web analytics methods can be applied successfully in a B2B application for a technology‐oriented company.Practical implicationsThe paper illustrates the use of clickstream data to measure the progression of web site visitors through the conversion process toward purchase.Originality/valueInsight is provided into the usefulness of web analytics as a framework for performance measurement that is used to drive success for B2B web sites.

Knuckles: bringing the database to the data

Click-stream data warehousing has emerged as a monumental information management and processing challenge for commercial enterprises. Traditional solutions based on commercial DBMS technology often suffer from poor scalability and large processing latencies. Although click-stream data tends to be collected in a distributed manner to support scaling the servers that host the websites, in general these partitioned click-stream logs are collated and pushed upstream to a centralised database storage repository, creating storage bottlenecks. In this paper, we propose a design of an ad-hoc retrieval system suitable for click-stream data warehouses, in which the data remains distributed and database queries are rewritten to be executed against the distributed data. The query rewrite does not involve any centralised control and is therefore highly scalable. The elimination of centralised control is achieved by supporting a restricted subset of SQL, which is sufficient for most click-stream data analysis. Evaluations conducted using both synthetic and real data establish the viability of this approach.

Development of a recommender system based on navigational and behavioral patterns of customers in e-commerce sites

In this article, a novel CF (collaborative filtering)-based recommender system is developed for e-commerce sites. Unlike the conventional approach in which only binary purchase data are used, the proposed approach analyzes the data captured from the navigational and behavioral patterns of customers, estimates the preference levels of a customer for the products which are clicked but not purchased, and CF is conducted using the preference levels for making recommendations. This also compares with the existing works on clickstream data analysis in which the navigational and behavioral patterns of customers are analyzed for simple relationships with the target variable. The effectiveness of the proposed approach is assessed using an experimental e-commerce site. It is found among other things that the proposed approach outperforms the conventional approach in almost all cases considered. The proposed approach is versatile and can be applied to a variety of e-commerce sites as long as the navigational and behavioral patterns of customers can be captured.

Association And Classification Models For Web Mining

We present methods for the search of association rules in the analysis of web clickstream data. We first show association rules, and show how to draw data mining conclusions on the basis of them. We then compare association rules, which are local models, with tree-based models, and show how association rules can be deduced from them. Our analysis have been conducted on a real ecommerce dataset.