Economic statistics are about the past. They are not what happened; at best they are an approximation to what happened. They are about indices and arbitrary measures constructed by humans to approximate and abbreviate a totality too large to be apprehended. Sometimes the necessarily limited nature of the statistics hides important events. The Consumer Price Index, for example, only features selected items and has to be updated to reflect changing lifestyles. Even then it can miss things not included in its list.
It is possible to study economic statistics and look for correlations between what might otherwise be seen as independent variables. Regression analysis can show that certain figures rise or fall together, or in inverse ratio. Equations can be constructed which show apparent relationships between some variables, and can give reasonably good approximations to the measured outcomes that have been recorded. This describes the past, however. It is when economic statistics are used to predict the future that problems arise. Most 'rational agent' models assume that people know past prices, and can form normally distributed expectations of future prices on that basis. But the future often defies those expectations.
From a room's doorway we can look at its contents, as we look at the past, gaining a reasonably good picture of what the room looks like. Even then, though, we are choosing what to notice out of vast numbers of things we could look at. When we turn our direction and step outside from that doorway we are stepping into the unknown, and the knowledge we take of what is behind us might be of little help in examining what lies ahead. The future is by its nature unknown and unknowable. Popper observed that since we cannot know now the content of future scientific discoveries, and since they will affect our future, we can never predict that future.
The phrase 'economic science' is misleading because economics deals with material not susceptible to the scientific method. We would think little of a scientific theory that changed its predictions "because people now know more than they did," or "because people now behave differently." Yet we are often asked to accept these as explanations of why an economic prediction failed, rather than accept that the theory itself was not up to the task.
We can, of course, conjecture relationships on the basis of past data, but these are always subject to modification or rejection on the basis of new outcomes. Economics deals with real people who all differ from each other, and who have individual motivations and social habits in a way that the material of the physical sciences do not. It has to deal with people whose behaviour changes in response to unpredictable events. The further that economic theory is removed from this world of motivated human beings, the less is it likely to tell us anything about the real world future their actions will bring about.