About us: Bokio is a free online service that covers accounting, billing, payments and expenses. We are committed to improving the everyday life for business owners by utilizing the power of artificial intelligence combined with great product design. 

We are looking for master students who want to do their master thesis project in machine learning. We have two possible projects to choose from. 

Project 1: Intelligent data analysis for anticipating customer churn

Description: Maintaining customer loyalty have been one concern in customer relationship management for many years. Customer churn is about the speed at which a customer leaves a company or service. Churn could happen due to many reasons and churn analysis can help to identify the cause and open the opportunity to design and implement retention strategies. The use of intelligent data analysis for the study of information can be of great assistance, this can give awareness and quantifiable metrics in churn management. The goal for this project at first is to analyze historical information to identify one or more patterns of correlation and establish the key factors for churning, after that to design a machine learning model that probabilistically identifies the segment when a customer is leaving the service, and finally evaluate the model. In preliminary research, it has been seen that related works to customer churn have been approached with classical machine learning techniques (support vector machines, k-means, decision trees, etc.); this can be an opportunity to explore the use of deep neural networks applied to this context.

Requested Experience:

  • Background in machine learning, experience in projects is a plus.
  • Good programming skills in Python, Matlab or R

References:

https://core.ac.uk/download/pdf/87651592.pdf

https://acadpubl.eu/jsi/2018-119-10/articles/10b/2.pdf

http://www.upv.es/deioac/Investigacion/ManuscriptDSS.pdf

https://arxiv.org/pdf/1712.08101.pdf

 

Project 2: Personalized predictive customer support

Description: Customers support is one of the main areas of interest for companies. This contributes to helping customers with answers about the functioning of a system or possible problems that may arise. Several means can be used to provide this information, one of the most conventional ways is by the customer itself searching through a web page for tutorials and videos. Sometimes this information can be extensive and difficult to manage, which makes it overwhelming. This opens the opportunity to develop new ways to present the information. The aim of this project is to analyze and understand historical interaction between the customer and Bokio application to predict and suggest possible inquiries. Two main challenges can be seen when it comes to finding a solution for this project. The first one is the selection of the historical information and its time frame, to obtain a meaningful and suitable data set. The second challenge is to determine the adequate model to make a good prediction. One idea to overcome both challenges is to work with ensemble learning techniques and define a specific goal for each entity. In that way, each entity could be capable to make suggestions analyzing the customer historical data, with different perspectives such as long-term historic interaction or the most recent contact with different parts of the application.

Requested Experience:

  • Background in machine learning, experience in projects is a plus.
  • Good programming skills in Python, Matlab or R

 

Interested?

Let us know which project you are interested in and why by sending your application to us before the 30th of June.

We are looking forward to hearing from you!

Apply for this job