I want to prepare my data for data mining, using a SOM to find data patterns. I have data about households in South Africa. The data is per individual per household. The census asked questions such as if the individual of a household was suffered any injuries or diseases in the last three months and if any what kind of injury or disease.

With this data, I can determine how many inhabitants of a province in South Africa contracted a disease as well as how many are injured. I can also see what the most likely disease or injury a household would be suffering per province and even get a general idea of the number of people injured or contracted a disease per age group per province. With this dataset, I want to find patterns provinces share but I don't know if I need to exclude or include the province as a one-hot encoded feature in my household data set or leave it out altogether.

I also tried to understand how to label provinces in a U-Matrix but most researchers explain what a U-Matrix is instead of actually explaining how they labeled their geographic feature inside a U-Matrix as seen here. Just point me to a keyword to research and I will do so.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.