The Deep Learning Revolution

To a greater or lesser extent, we business, finance and real estate researchers are all empiricists, making a presumption that knowledge is based on experience and that data is essential to prove any hypothesis we might form about the way the world works. Of course, we know that this is flawed. Journals are full of papers that claim generalisations, formed from samples of data or experiences that are very specific to locations or time periods. We will use the present tense and claim that A causes B with 87% explanatory power, when we should say that A appears to have caused B over the past 15 years in the USA. Most of us will feel much better about making such generalisations if the finding supports, or is supported by, a normative theory to the effect that A should cause B. Usually, empirical tests are used to prove or disprove a prior (hence a priori) theory. Theory precedes empirical tests.