Data Challenge

  • Global Case - Raiffeisen Bank
    Case Name:
    Cold start Predictive modeling for NTB customers

    Problem Description:
    Understanding the customers and their behavior is crucial for offering them a superior customer experience. Collecting relevant information from different sources is something that new cutting-edge technologies and tools enable and there are always some untraditional sources that are not regularly considered.

    In recent years, numerous emerging open data sources evolved and can be used with an aim to enrich the existing information about the customer. This open data can be assessed and combined with existing information from core banking systems. Given consent, customer behavior and online digital footprint can help better understand their unique profile and similarities with other customers.

    This is especially important when it comes to new-to-Bank customers (completely new customers from the Bank perspective). For such customers, the Bank does not have enough information about their needs, and it is very hard to offer the right products and services in the early months. Therefore, the customer is losing an opportunity to receive products fitting their needs and must wait until a sufficient amount of information is in the system so that personal preferences can be evaluated.

    Scarce customer information from the early stage of relationship with the Bank (personal information, addresses, contact information, etc.) can be enriched with online open information (geolocations, maps, and places, web searches, official public statistics, social feeds, etc.). This enables the Bank to build ML models that predict probabilities that the customer could benefit from a particular product/service, either by comparing them to similar existing customers or by finding meaningful patterns in their online behavior and presence. This could significantly improve customer satisfaction from the very beginning of their relationship with the Bank.

    The bank is providing historical information about customers' behavior during their first months of relationship with the Bank (product usage, transactions, balances, finance data, segmentation, etc.).
  • Regular Case - VIP
    Case Name:
    Handset Recommender

    Problem Description:
    We as a TELCO company are eager to satisfy our customers. We are curious about their needs.

    For our existing customers, we know a lot about their behavior during our relationship but for external customers, we know just a few basic information about their attitude towards expenses and brand preferences. We would like to offer them a handset that is most relevant, using rule-based internal prioritization but also in parallel to offer them handsets that are based on hype at the market.

    For this purpose, we will prepare data sets of historical data for every existing customer: handsets bought in last few years, handsets used in last few years, tenure of the customer, handset characteristics (brand/technology2g/3g/4g/price…), periodicity of handset changes, payment behavior, usage behavior, etc.

    For external customers will have just a few data (brand preference, price ready to pay) but we are willing to understand them as well. For internal portfolio prioritization we will also deliver files, but to find “hype” handsets you should investigate on the market place also (YouTube/Google/GSM Arena/Social media …). You can use various open-source machine learning and statistical tools that can help you in pattern detection.

    Are you ready to dive into our data and help us to make our customers delighted with our offer?

    Join us and let's have fun!
  • Regular Case - UNDP
    Case Name:
    Better access to inclusive education

    Problem Description:
    Inclusive education was introduced as part of the regular schooling system in order to provide the best possible learning environment for children with disabilities. Inclusive education means that all the obstacles for learning (physical and communicational) in and around schools need to be removed. Individual learning plan (ILP) is developed as part of inclusive education and defined within a legal framework (Law on the fundamentals of the education system). The goal of ILP is the optimal development of the child and pupil, inclusion in peer collective and achievement of general and specific learning outputs.

    The Ministry of Education, Science, and Technological Development released open data on teaching institutions in Serbia, including the number of ILPs in every grade (ILP 1, ILP 2 and ILP 3 – from basic to upper level). The access to this data is important in order to provide relevant information for parents about those educational institutions which have more ILPs developed. The greater number of ILPs shows better expertise of named school in inclusive education, and better learning opportunities for their children.

    In order to make this data available for the parents and education policy agents, there is a need to show this data in an appealing way to the target groups (map, dashboard, a searchable database including the location of the school, etc.). Applicants are invited to use any useful open data set, apart from the recommended data.

    Join us in making open data work in favor of a better life for people!


Applications are closed


Engineers, Analyst and Data Scientist from the industry


Researchers and university representatives


Students with advanced knowledge of Data Analytics

Don’t miss your chance to compete with data scientists from 4 different European countries.
Applications are closed. The number of teams is limited per host country, so hurry up and apply. Only teams from the host country may apply.

Important dates you need to remember:

Full case and dataset reveal: October 3
Datathon: October 12-13
Results: October 27

Teams need to consist of 3 to 5 members. The participation fee is 100€ per team.