Srijeet Chatterjee

Srijeet Chatterjee

Master's Thesis

Data-Driven Customer Experience Management

Advisors
An Nguyen (M.Sc.), Prof. Dr. Björn Eskofier

Duration
10/2019 – 05/2020

Abstract
The cutting-edge medical imagining software helps to identify diseases from images of different modalities like CT, MRI and others on a single platform. Clinics are exceedingly dependent on them to conduct tasks of their routine clinical workow. Maintenance activities like upgrades and
updates of the software and time-consuming waiting periods during failures are inconvenient for hospitals. Hospitals need a fast solution to such technical hitches and comprehensive solutions from the provider’s end. The service priority is to deliver the best interaction which improves
the overall customer experience [1] and ensures the availability of systems for the patient. Today companies like Google focus on the importance of user experience research and data-driven administrative decisions[2]. The target recommendations for this project are the best sequence of service activities for the customer based on existing customer service data and the most probable next sequence of service activities[3]. Next, predictions for proactive servicing and the estimated time for overall service activity[4] are also targeted.

The goal of the thesis is to improve the customer experience through personalized service recommendations. The research investigates whether process mining and deep learning methods help to achieve better results compared to traditional prediction methods. Finally, the validation of the
methods is carried out and results are discussed.

The various steps required for the development of the project can be listed as follows:

  • Data preparation: Information regarding customer service is collected. The data should reect current service activities including interactions with the customer. The customer journey data is to be prepared such that it includes the attributes: case-ID, activity and time-stamp which are essential for further analysis. Additional context information will be collected as well.
  • Process Mining: The process discovery models [5] are applied to understand the current customer journey process.
  • Recommendations: Tailor-made approaches are sensitive to dataset changes. Deep Learning based methods are robust regarding the prediction of processes[6]. In this project, the customer journey process is recommended. Furthermore, a variety of common approaches like collaborative and content-based [8] recommender systems will be implemented for comparison.
  • Validation: Methods must be validated and benchmarked with existing methods. The results obtained must be discussed in detail. The deep learning-based models exploit techniques like sequence embedding for ecffiently estimating vector space distance between two service activities to predict activities that are highly likely to happen in a sequence [7]. Next, recurrent architectures like Long Short Term Memory(LSTM) [6] are applied for predictions. Finally, the methods should be validated.

References:

  1. von Zernichow, Roger, Marita Skjuve, and Ragnhild Halvorsrud. Customer Journey Heatmaps: a wake-up call. Proceedings of the 10th Nordic Conference on Human-Computer Interaction. ACM, 2018. https://doi.org/10.1145/3240167.3240277.
  2. Au, Irene, et al. User experience at google: focus on the user and all else will follow. CHI’08 Extended Abstracts on Human Factors in Computing Systems. ACM, 2008. https://doi.org/10.1145/1358628.1358912
  3. Huber, Sebastian, Marian Fietta, and Sebastian Hof. Next step recommendation and prediction based on process mining in adaptive case management. Proceedings of the 7th International Conference on Subject-Oriented Business Process Management. ACM, 2015. https:/doi.org10.1145/2723839.2723842.
  4. Bolt, Alfredo, and Marcos SepÃo lveda. Process remaining time prediction using query catalogs. International Conference on Business Process Management. Springer, Cham, 2013. https://doi.org/10.1007/978-3-319-06257-0_5.
  5. Van Der Aalst, Wil. Data science in action. Process Mining. Springer, Berlin, Heidelberg, 2016. 3-23. https://doi.org/10.1007/978-3-662-49851-4_1.
  6. Tax, Niek, et al. Predictive business process monitoring with LSTM neural networks. International Conference on Advanced Information Systems Engineering. Springer, Cham, 2017. https://arxiv.org/abs/1612.02130.
  7. Mikolov, Tomas, et al. Ecient estimation of word representations in vector space. arXiv preprint arXiv:1301.3781 (2013).https://arxiv.org/abs/1301.3781.
  8. Portugal, Ivens, Paulo Alencar, and Donald Cowan. The use of machine learning algorithms in recommender systems: A systematic review. Expert Systems with Applications 97 (2018): 205-227. https://doi.org/10.1016/j.eswa.2017.12.020.