Challenge

 

Is it possible to relate a Sustainable Development Goal (SDG) with policy indicators? If so, one could determine what is needed to achieve a SDG.

Select from a SDG, one indicator, and collect data. Use the three machine learning techniques to predict the selected indicator. Write a report with the findings, with not only the differences between ML but also with the most important factors that affect the selected indicator, and in what way. 

• Big idea: use country data over the years do identify what is important to a SDG
• Essential question: can an indicator of a SDG be learned by a ML, if available relevant macro country data?
  
• Challenge: select one indicator from one SDG, collect data related to the indicator, and try to use three machine learning techniques to predict the indicator
• Guiding questions:  
  • Exist data on countries, with indicators on both policy and SDG status, over the years?
  • Why not use machine learning methods to relate SDG with policy indicators?
  • It is possible for a specific SDG indicator, to know what is important?
• Guiding activities, resources: 
  • Task 1 / 2
  • Country data in: https://www.pordata.pt/en/home
• Solution: open challenge, each participant investigate its own indicator, and share with others
  
• Evaluation and sharing: participants submit the report in the forum, and give comments on others reports.
  

Resources:

Data: PORDATA - Statistics, graphics and indicators

Explore existing data in the PORDATA website.

Choose one of the sustainable development goals (Europe Database | Pordata), and one of the related indicators, and extract data to test the methods for predicting that indicator on the basis of the other indicators collected. Alternatively, use any other indicator that you consider relevant, and try to obtain a relationship between that indicator and other indicators that you consider to be critical.

   

You can use chatGPT to find out what the indicator in question depends on. If that indicator really depends on the ones listed, if data is available, then a learning method will be able to learn the value of the indicator under study based on the indicators it depends on. Check that this is the case, and analyze the results of the methods for confirmation.

You can consider several years as separate observations, but to ensure independence between cases, don't use consecutive years.

Before you go any further, check in the forum that no one is trying to learn exactly the same indicator.
Then post in the forum the domain, the indicator you want to learn based on a set of others on which you suspect there is a dependency.
Share your findings in the end.