Interactive Event Sequence Prediction for Marketing Analysts (original) (raw)

Survey on Visual Analysis of Event Sequence Data

arXiv (Cornell University), 2020

Event sequence data record series of discrete events in the time order of occurrence. They are commonly observed in a variety of applications ranging from electronic health records to network logs, with the characteristics of large-scale, high-dimensional and heterogeneous. This high complexity of event sequence data makes it difficult for analysts to manually explore and find patterns, resulting in ever-increasing needs for computational and perceptual aids from visual analytics techniques to extract and communicate insights from event sequence datasets. In this paper, we review the state-of-the-art visual analytics approaches, characterize them with our proposed design space, and categorize them based on analytical tasks and applications. From our review of relevant literature, we have also identified several remaining research challenges and future research opportunities.

A Visual Analytics Approach to Comparing Cohorts of Event Sequences

2016

Title of dissertation: A VISUAL ANALYTICS APPROACH TO COMPARING COHORTS OF EVENT SEQUENCES Sana Malik, Doctor of Philosophy, 2016 Dissertation directed by: Professor Ben Shneiderman Department of Computer Science Sequences of timestamped events are currently being generated across nearly every domain of data analytics, from e-commerce web logging to electronic health records used by doctors and medical researchers. Every day, this data type is reviewed by humans who apply statistical tests, hoping to learn everything they can about how these processes work, why they break, and how they can be improved upon. To further uncover how these processes work the way they do, researchers often compare two groups, or cohorts, of event sequences to find the differences and similarities between outcomes and processes. With temporal event sequence data, this task is complex because of the variety of ways single events and sequences of events can differ between the two cohorts of records: the str...

Guest Editorial: Multimedia for Predictive Analytics

Multimedia Tools and Applications

Recent technological advancements have led to a deluge of multimedia data from distinctivedomains. With the increase in multimedia data generated by Internet, health care, scientificsensors, financial and manufacturing companies has profoundly transformed our society andwill continue to attract diverse attentions from both technological experts and the society ingeneral. The use of multimedia data for predictive analytics is the process of analyzing,meaningful patterns for predictive modeling. Reasons for using predictive analytics

Simplifying Overviews of Temporal Event Sequences

Proceedings of the 2016 CHI Conference Extended Abstracts on Human Factors in Computing Systems, 2016

Beginning the analysis of new data is often difficult as modern datasets can be overwhelmingly large. With visual analytics in particular, displays of large datasets quickly become crowded and unclear. Through observing the practices of analysts working with the event sequence visualization tool EventFlow, we identified three techniques to reduce initial visual complexity by reducing the number of event categories resulting in a simplified overview. For novice users, we suggest an initial pair of event categories to display. For advanced users, we provide six ranking metrics and display all pairs in a ranked list. Finally, we present the Event Category Matrix (ECM), which simultaneously displays overviews of every event category pair. In this work, we report on the development of these techniques through two formative usability studies and the improvements made as a result. The goal of our work is to investigate strategies that help users overcome the challenges associated with initial visual complexity and to motivate the use of simplified overviews in temporal event sequence analysis.

Decision Exploration Lab: A Visual Analytics Solution for Decision Management

We present a visual analytics solution designed to address prevalent issues in the area of Operational Decision Management (ODM). In ODM, which has its roots in Artificial Intelligence (Expert Systems) and Management Science, it is increasingly important to align business decisions with business goals. In our work, we consider decision models (executable models of the business domain) as ontologies that describe the business domain, and production rules that describe the business logic of decisions to be made over this ontology. Executing a decision model produces an accumulation of decisions made over time for individual cases. We are interested, first, to get insight in the decision logic and the accumulated facts by themselves. Secondly and more importantly, we want to see how the accumulated facts reveal potential divergences between the reality as captured by the decision model, and the reality as captured by the executed decisions. We illustrate the motivation, added value for visual analytics, and our proposed solution and tooling through a business case from the car insurance industry.

ICE: Identify and Compare Event Sequence Sets through Multi-Scale Matrix and Unit Visualizations

ArXiv, 2020

Comparative analysis of event sequence data is essential in many application domains, such as website design and medical care. However, analysts often face two challenges: they may not always know which sets of event sequences in the data are useful to compare, and the comparison needs to be achieved at different granularity, due to the volume and complexity of the data. This paper presents, ICE, an interactive visualization that allows analysts to explore an event sequence dataset, and identify promising sets of event sequences to compare at both the pattern and sequence levels. More specifically, ICE incorporates a multi-level matrix-based visualization for browsing the entire dataset based on the prefixes and suffixes of sequences. To support comparison at multiple levels, ICE employs the unit visualization technique, and we further explore the design space of unit visualizations for event sequence comparison tasks. Finally, we demonstrate the effectiveness of ICE with three real...

Developing Effective Tools for Predictive Analytics and Informed Decisions

Technical Report, University of Tallinn, 2013., 2013

By utilizing the statistical analysis, analytics, information processing and business intelligence the business processes are understood and decisions are made aiming to improve profitability. Yet due to the involvement of big data, highly non-linear and multicriteria nature of decision making scenarios in today’s governance programs the complex analytics models create significant business, operational and technology risks as well as modeling errors presenting the lack of effective modeling system to governance programs. Consequently the traditional approaches have been reported less useful in proper guiding decision-making communication and in drawing insights from big data. Alternatively here the proposed methodology of integration of data mining, modeling and interactive decision-making is studied as an effective approach where what-if scenarios are evaluated and optimization-based decisions are made.

Forecasting software visualizations: An explorative study

A qualitative explorative evaluation considered the effects of six visualization interfaces of sales forecasting systems on 60 university students. The study builds on earlier research from the domain of business forecasting in supply chain industries. The evaluation generates exemplar interfaces derived from the theoretical framework and task analysis of interviews with 20 expert users and designers of forecasting systems. The implications for information visualization and interaction design are discussed.

A Visual Analytics Technique to Compare the Performance of Predictive Models

Advanced Visual Interfaces. Supporting Artificial Intelligence and Big Data Applications, 2021

Decisions that people make every day are affected by the information available in a given moment. Predictive models are used to estimate future values. For a given set of data and an analysis goal, the results of the models can vary, so it is important to select the most accurate model for the set of data. This paper proposes a Visual Analytics technique for comparing the performance of predictive models. It consists of four main components that support the tasks of the Keim's Visual Analytics Mantra: "analyze first, show the important, zoom, filter and analyze further, details on demand". The first component, analyze data, by building predictive models using various machine learning algorithms; the other three components are interactive visualizations that show the important results found by the models, zoom and filter on results of interest and finally, further analyze the selected results by showing details on the data.

Predictive analytics in customer behavior: Anticipating trends and preferences

Results in Control and Optimization, 2024

In order to effectively manage their customers, businesses need to thoroughly analyze the costs and advantages associated with various alternative expenditures and investments and determine the most effective way to allocate resources to marketing and sales activities over time. Those in charge of making decisions will reap the benefits of decision support models that estimate the value of the customer portfolio and tie expenses to customers' purchasing behavior. In the current work, various machine learning algorithms such as Decision Tree (DT), Random Forest (RT), Logistic Regression (LR), Support Vector Machines (SVM), and gradient boosting are used to predict customer behavior. The evaluation criteria considered in the work include precision, recall, F1-Score, and ROC-AUC. The accuracy values obtained for DT, RT, LR, SVM, and gradient boosting are 0.787, 0.806, 0.826, 0.826, and 0.823, respectively. The results emphasize RT and LR's good performance, while the values of 0.620, 1, 0.766, and 0.878 for the precision, recall, F1-score, and ROC-AUC score outperform the rest. The novelty of this work lies in employing a comprehensive set of machine learning algorithms to predict customer behavior, with a particular emphasis on the superior performance of RF and LR models, as demonstrated by their high precision, recall, F1-score, and ROC-AUC values.