Current Experiments

Data 4 Mobility:
Assessing the Potential of New Data Resources for Urban Mobility Planning

EXPERIMENT 1
P4 Testing
Potential countries: Tanzania, Thailand

By testing the use of non-traditional data sources, we want to discover new ways to build inclusive and sustainable transport systems. For this purpose we conduct experiments using and combining different types of data. In Bangkok, Thailand, we are experimenting with data from ride hailing services to find out if and to what extent we can gain insights on the urban transport system from the data. In Dar es Salaam, Tanzania, we are building a prototype of an integrated platform that combines different data sources to help decision-makers gain actionable insights into urban transport routes.

Read more about

AI Training Data for Agriculture: An Experiment in Artificial Intelligence “Learning”

EXPERIMENT 2
P4 Testing
Potential countries: Burkina Faso

In this experiment we test whether we can use our ‘internal data treasure’ to train artificial intelligence models that may serve to improve our services for our target group/project beneficiaries. In a first step we focus on determining how data gathered from GIZ projects in the agricultural sector may be used to train sophisticated crop yield projection models.

Read more about

THE CASE FOR DATA MERGING: GATHERING ACCURATE DATA IN FRAGILE CONTEXTS

EXPERIMENT 3
P1 Understand
Potential countries: Afghanistan, Somalia, Yemen

The lack of quantitative, reliable and up-to-date data is a major obstacle for evidence-based, high-quality service provision in fragile – frequently inaccessible, target areas. In this experiment the Data Lab will focus on healthcare as one of the most promising sample sectors and in order to try to increase the validity of service coverage projection through the combination of qualitative and quantitative data. This will help to better plan the maintenance of existing healthcare infrastructure and (to a lesser extent) to plan the development of new healthcare facilities.

Read more about

Data Powered
Positive Deviance (DPPD)

Experiment 4
P3 Prototype
Potential countries: Indonesia, Mexico, Nepal, Niger, Somalia

Using large datasets, such as satellite data or online search information, we are trying to identify positive outliers, often referred to as positive deviants, that fare better than their peers with regards to crucial development outcomes. Once identified, the aim is to understand and later to scale their locally-sourced solutions within their communities.

Read more about

POTENTIAL EXPERIMENT TOPICS
APPLICATION IN DIFFERENT SECTORS

  • Crowd Sourcing

    Citizens and service recipients participate in the collection of data

  • GIZ ‚dataset-treasure‘

    Relevant tools to process the data of development projects for (strategic) decision making

  • SDG Tracking/M&E

    M&E with high quality data at micro, meso and macro level

  • Data privacy & -security

    How to protect data privacy and - security rights while working with Big Data (analytics)

  • Open Data

    Public availability of data across organizational boundaries for increased transparency and synergies

  • AI Ethics

    How to ensure that outcomes of AI applications are just and do no harm