ETH students assist NGOs in war zones

We talked to Simon Mathis and Renato Durrer, co-founders of the Analytics Club at ETH.

Thousands of people living in countries like Syria suffer terribly from the war and depend on support from outside. How do affected people receive help?

Living conditions in war zones can change daily, which means that the availability and cost of goods is unpredictable. Hence, rather than distributing goods, which can be dangerous and a logistical challenge, organisations such as the World Food Programme are moving towards direct cash transfers. This is often more efficient and cost-effective than traditional forms of aid. To be effective, direct cash transfers require an accurate estimate of the price of goods needed to survive in a certain region. At the global level, organisations like our partner NGO IMPACT Initiatives conduct regular Market Monitoring exercises to inform 2.8 billion USD of global humanitarian spending on cash and voucher assistance.

What are the difficulties of direct cash assistance to crisis-affected populations?

To enable effective cash distribution, NGOs must know if basic survival items are available to buy and what the cost of these items is. They determine the cost by market monitoring and the calculation of an SMEB (Survival Minimum Expenditure Basket). Since the SMEB is composed of many items, it is difficult to calculate in case of missing items due to shortages. Any new outbreak of an armed conflict prohibits data-collection in such regions and increases the problem of missing data. Furthermore, prices vary enormously depending on time and location. To solve the imputation problem when data on some items are unavailable, IMPACT was looking for a reliable algorithm to predict the potential values of unavailable items.

Therefore, the task for the Analytics Club at ETH was to create a reliable algorithm.

It was indeed. A challenge we happily accepted for our Hack4Good programme. Hack4Good matches data science talents from ETH Zurich with non-profit organisations that work in service of humanity. Partner organisations benefit from having access to the ETH talent pool and student teams working free of charge.

Over a period of two months, our Hack4Good teams of four students each worked hard to develop a reliable and useful algorithm to calculate the SMEB, even in the presence of some missing data. The intense work of many dedicated students provided a solution that convinced IMPACT that these solutions should be further developed and eventually deployed for their market monitoring analysis in Syria and beyond.

What options do companies have, if they are looking for expertise and support in the field of machine learning?

Depending on their needs, there are several options:

• They can offer 1-2 days ideation projects at the Analytics Club at ETH Datathon
• For non-profit organisations the Hack4Good programme is a good option
• Smaller projects can be handled by the ETH Juniors
• For larger projects of up to 12 months duration there are start-ups such as Visium, which is co-founded by ETH Alumni and offers tailored artificial intelligence solutions for companies
• For extensive research projects, there is the possibility to work with ETH professors or the Swiss Data Science Center.

What are your next moves?

Having graduated from ETH, we will now hand over the Analytics Club to a new team and concentrate on the development of the artificial intelligence start-up Visium. At Visium, we support clients in their machine-learning projects. Our aim is to close the gap between the latest research results and industrial applications.