DATATHON Spring 2025

Sign up now!

Key Facts of past Datathons

100 +

Participants

7

Challenges

1000 €

Prize money

What happened at the STADS Datathon?

Participants anticipated two days of intense teamwork focused on data science, data visualization, or programming challenges, along with plenty of fun.

The benefits included:

🚀 €1,000 for the winning team
🚀 Exciting prizes
🚀 Food + drinks included
🚀 Get-together with many companies

 

 

Where? Mannheim – CUBEX ONE!
When? On 07.-08. of March 2025!

The Timeline

An insight into past challenges

Ernst & Young

Thousands of payments are processed monthly by large companies. It is essential to ensure that each invoice receives exactly one payment. Issues with data quality, duplicate invoices, or attempted fraud against the company make this a challenging and complex problem in the daily work of auditing.

In this challenge, the goal is to identify duplicate payments for the same invoice within a large dataset. You can choose from various approaches to solve this problem. We are excited to see what ideas you come up with.

PHOENIX group

The PHOENIX group connects the medical aspect of its business with the currently highly relevant topic of ChatGPT and generative AI in this challenge. We challenge you to build your own chatbot in a short period of time. Based on ChatGPT from OpenAI, you should create a symptom checker. It should have a functional interface and demonstrate real expertise. Interested?

msg systems

The Challenge is about predicting Price shocks.

A price shock is a sudden surge in material prices, which can lead to rising production costs and even to an uneconomical production. In order to be able to predict these shocks, a model was trained using key factors that have caused sudden increases in prices in the past. The model can predict future prices and identify possible shocks based on currently occurring key factors. To provide these current factors to the model as features, they must be extracted from current news articles. 

 

RSM Ebner Stolz

The year-end puzzle: Identify the entries that are decisive for the account balance with RSM Ebner Stolz.

One of the main tasks of auditors, for example, is to assess the valuation of a balance sheet item at the end of the fiscal year. The amount of the balance sheet items to be valued depends on the business transactions on the underlying accounts of the company. They are recorded in the form of entries. Over the course of the fiscal year, for example, assets such as receivables are recorded in an account and derecognized when they are settled. In many cases, there is a so-called sub-ledger that can provide information about the composition of an account, but this does not apply to all accounts. In the midst of the greatest stress, the audit manager needs this information for an account, but the accounting department is already completely overwhelmed with other issues. Can you use your analytical skills and algorithmic thinking to crack this numerical puzzle?

 

Ebner Stolz

Auditing uses financial metrics such as equity ratio and return on sales to assess company performance and health. Benchmarking, or comparing metrics between companies or over time, facilitates the audit of the financial statements. Anomalies should be investigated when metrics significantly deviate from benchmark values. But what deviations are considered significant? When does benchmarking identify anomalies that the audit cannot ignore? Ebner Stolz provides an anonymized dataset and offers support for anomaly detection. The choice of comparison perspective is crucial for success.

BearingPoint

The assessment, processing, and settlement of claims are core business activities and cost drivers for insurance companies. Today, assessments are primarily conducted based on image data and on-site by experts. This process is error-prone, time-consuming, and highly manual. A faster, automated, and more accurate assessment of damage amounts represents a significant improvement for companies and employees. You are to support the AI-based estimation of damage amounts with the additional use of computer vision methods! You are free to choose the methods, approaches, and models you apply. A dataset containing image material, label information, and other features related to water damage will be provided to you.

STADS

McCar, a fictitious used car dealer from Mannheim, always offers is customers the best prices for their new dream car.

Due to constant growth in recent years, McCar unfortunately no longer keeps up with the pricing. Everyday, the optimal sales prices has to be found for dozens of vehicles.

McCar has a data set with the characteristics and sales prices of the last year at its disposal – can you support them and develop a model that predicts the optimal sales price for every vehicle in the future?

 

We thank our Sponsors und Premium Partnerns

Logo unseres Kooperationspartners zeb
Logo unseres Kooperationspartners zeb
Logo unseres Kooperationspartners zeb
Logo unseres Kooperationspartners zeb

We thank our Sponsors und Premium Partnerns

Logo unseres Kooperationspartners zeb
Logo unseres Kooperationspartners zeb
Logo unseres Kooperationspartners zeb
Logo unseres Kooperationspartners zeb

Don’t hesitate to contact us!

Apollon Karalis

Datathon team lead

Bild des Kooperation Teamleiters Konrad Walter