Why do we need to distinguish between strong and weak correlation of data sets using best-fit lines? When a correlation has been established between the two data sets, how is it linked to causality? Please provide examples to support your answer.