To predict the popularity of a house, there might be several factors that could be relevant, like size, price, distance to mrt - while some others can be difficult to decide, like the education of current owner etc.
How do we know whether the feature is useful in predicting the popularity of a house?
What I could think of head is correlation analysis. But is it possible that it shows no correlation, but in the end, it ends up that the correlation only manifests when consider several features together?
