(1) Collect data.
(2) Analyze the data, looking for a trend. (If found, the trend is a "Natural Law*.")
(3) Hypothesize a reason for the trend.
(4) Test the hypothesis. If it fails, repeat (3) as many times as necessary.
(5) If the scientific community agrees, the hypothesis is accepted as a theory (with your name on it, if you are lucky).
These days, steps (1) and (2) can be done by a computer. The problem, of course, is that people can look at a set of data and never see a trend, and, if they see a trend, never form a valid hypothesis. (I liken the scientist forming a hypothesis to an artist creating a work of art from a blank canvas.)
*A Natural Law does not need an explanation. You can create a cuckoo clock driven by a falling weight without knowing anything about Newton's Theory of Gravitation.
