Community

Hi Tamsin
As I'm describing in the article, the key is to use various data anonymization techniques. Sometimes, you change the original data so that confidential and sensitive information is lost. Still, the context remains (you don't need to know the exact payment location, but the city is enough for the model). Sometimes, you want to create synthetic data that are not using any real information but allow the model training. And sometimes, you just exclude the whole set of data features, knowing that the model doesn't need it at all.