Sky is where the view is, dear Kamal. The view is always from above. Whether empirical or experimental, dear Antonio, the method is not much of a question to me. The result(s) is/are!

I think that the extensive use of empirical correlations and empirical methods can be more applicable for studies in Humanities. These methods are less used in context of the current scientific developments and in the Engineering practice. 

Thanks Kamal and Marwan for so intelligent answers. I must confess that agree a little more with Kamal (solid ground), but thanks to God that Marwan has also reason in many cases.

There is a new approach of research through practice and depends on empirical methods, in which, practice knowledge (empirical methods) take place, as a main resource, instead of theoretical knowledge, especially in architecture.

Both, theoretical and empirical methods are scientific approaches. 

Deletion methods which are designed to apply to analogy-based estimation is an empirical one.

Existing outlier deletion methods

• Cook's distance based deletion,
• LTS based deletion,
• k-means based deletion,
• Mantel's correlation based deletion, and
• EID based deletion
• Analogy-based estimation.

Datasets

• ISBSG,
• Kitchenham, and
• Desharnais datasets

Estimation accuracy evaluated through

• BRE (Balanced Relative Error).
• Eliminates outliers from the neighborhoods of a target project when the effort is extremely different from other neighborhoods.

http://dl.acm.org/citation.cfm?id=2020407

The differences are rooted in their background education (secondary or high school) whether they are familiar with (studied) maths or  science. For instance, most of agricultural sciences (agronomy, horticulture, ....) prefer empirical methods whereas agricultural engineering (irrigation, machinery, etc.) use scientific approaches as well as computational intelligence nowadays.

Definitely, dear Kamal. There are thirteen ways of looking at the bird--figuratively speaking!

One such empirical method using Technology applied in practice.

Register complaints via video conference at Bangalore Police's first remote FIR centre

If you feel intimidated at the sight of a police station, you can avoid going to Bangalore. In the Silicon Valley of India, one can just walk to a kiosk known as the Remote FIR Registration Centre, pick up a phone and register a complaint through video conference. It is a 24/7 centre and has the look of a plush Bank ATM.

Once the complainant contacts the investigating officer via video conference, the concerned police officer at the control room will register the case and a copy of the FIR will be delivered on the spot.

The city police has opened two FIR registration centres on an experimental basis in Bangalore. Chief Minister Siddaramaiah inaugurated the two centres on Friday and praised the police for embracing the modern technology with open arms.

http://www.msn.com/en-in/money/technology/register-complaints-via-video-conference-at-bangalore-polices-first-remote-fir-centre/ar-BBdEuwm?ocid=mailsignout

Let’s give an example in medicine (I have several medical doctors in my family and a profound respect for the medical doctors job and its noble mission, and I know that they do always the best they can):

When my son was a baby (21 years ago) he always slept with his belly down. When my daughter was a baby (9 years ago) she slept always with the belly up and the bed leaning. Currently I don't know what the empirical correlation advised with respect to this issue is.

Empiricism is still very common (and probably useful?) in all knowledge areas.

Thank you dear Mahmoud for remember us the importance of the secondary school in our formation (probably I was lucky to live in a little town that has one of the oldest universities of Europe). Some basic Philosophy concepts and specially the Hegel dialectic are still always present in me. Probably, the scientific method (at least) and eventually some basic Philosophy concepts (or the Hegel dialectic) should be always present when we read or do research. I let here what the Scientific Method is according to the Oxford dictionary:

“A method of procedure that has characterizednatural science since the 17th century, consisting in systematicobservation, measurement, and experiment, and the formulation, testing, and modification of hypotheses:criticism is the backbone of the scientific method”

http://www.oxforddictionaries.com/definition/english/scientific-method

Theoretical approach is highly preferable in engineering research. However they involve assumptions in mathematical modeling requiring knowledge of numerical procedures for obtaining solutions. Experimental results are applicable in the range of operating parameters. A combination of theory and experiments can give insights to the physics of the problem.

Dear António, by scientific method we mean the following 4 steps:

Dear Sharma. Very good explanation. I always think of the former (Theoretical approach) as deductive approach But generation of new theory emerging from data (Experimental results) as inductive approach. Accordingly, deduction begins with a universal view of a situation and works back to the particulars; while induction moves from fragmentary details to a connected view of a situation.

From my point of view correlations for momentum,heat ,and mass transfer will be needed whenever technical problems (often complex in nature) have to be solved at least approximately. The art of modelling is to reduce these complex problems to a combination of standard partial problems for which approximate solutions in terms of correlations exist.

It is immideately obvious that there is the need for as many correlations as possible that can solve certain partial problems. They, however, can be obtained completely empirically, or at least to a certain extent based on theoretical considerations, as for example to determine the entropy generation when losses in a flow field should be described in terms of a certain correlation.As soon as turbulence is involved there will be always an empirical part in the determination of correlations.

Thanks for the excellent answers, dear Karma and dear Mahmoud (of course those little details can be always discussed). Thanks dear Eraldo for expressing also so clearly your point of view in this matter.

Thanks Dear Heinz for sharing with so much simplicity your profound knowledge in the field of thermodynamics. Many water flows in nature and in the most of practical applications occur typically in turbulent flow. So (at least) for water turbulent flows, I agree with you. The current research requires a semi-empirical basis, as you said:“As soon as turbulence is involved there will be always an empirical part in the determination of correlations.”

I don't think that any field can be developed without theoretical approach even the empirical one. More than 100 years ago infinities computations were not useful as thought by many scientific researchers. However, its implication came late to the computer evolution. Assumptions are always there in all kind of research with certain kind of limitation and delimitation. This is a good reason to replicate research method  by using different technique.

Thank you for your excellent answer, Dear Hashem. Your peculiar way of thinking is challenging and a good training for brains. Generically speaking I agree with you. Numerical analysis allows us currently the numerical integration of very complex differential equations, but some modelling of some real phenomena is really complex. On the opposite extreme, theoretical assumptions in empirical methods sometimes exist, even if sometimes limited.

An exception to the anticonsciousness zeitgeist was the work of Ryan. Anticipating the cognitive revolution in psychology, It seems a simple fact that human behavior is affected by conscious purposes, plans, intentions, tasks and the like" (p. 18), which he called first-level explanatory concepts, were the immediate motivational causes of most human action.

http://faculty.washington.edu/janegf/goalsetting.html

I do not understand following sentence: The extensive use of empirical correlations and empirical methods instead of scientific approaches can be controversial.

Empirical approaches (e.g.observation) belong to scientific approaches. Without observation, no human-based science

Thanks dear Marcel for your contribution. Scientific approaches (based on the scientific method) and fully empirical approaches based frequently on mere correlations are quite distinct, even if both normally use laboratorial or field experiments, computational results, or other observed data. For example in optimization, I can clearly distinguish a hill climbing algorithm (empirical) and a simulated annealing (scientifically based, resulting of the mimic of an observed natural process). Formulas or methods can be merely empirical, semi-empirical, or fully scientific based. The last normally uses also experiments, observation and measurement of real phenomena or computational results. I would not call empirical data or heuristics to the results of experiences performed according to the scientific method. In my opinion, words like empirical and heuristics (depreciatory meaning) are probably used many times out of the suitable context.

Dear all. Any person (an empiricist, or theorist or practitioner with no education or a qualified practitioner ...) before starting physical work initially designed stages of action. Each takes into account previously acquired knowledge. At the first stage, each acts as a theorist. Then, as a practitioner. If the first stage will not be, the result is not very good.

I propose to take into account the second argument. Often humanity must solve complex problems. The solution of such problems in the short deliberation impossible. For example, it was during the initial knowledge of nuclear energy. How much did the theoretical physicists in the direction of the knowledge of this phenomenon.

But most importantly, the success of each activity reaches a professional. I support professionals theorists and practitioners.

I agree with @ Heinz Herwi, to determine the entropy generation when losses in a flow field should be described in terms of a certain correlation.As soon as turbulence is involved there will be always an empirical part in the determination of correlations

Thank you for expressing your agreement, Dear Krishnan. In fact (in my opinion) both K. V Sharma and Heinz Herwig answers are excellent (deep knowledge in the respective research areas).Let me remember some sentences of them: “Theoretical approach is highly preferable in engineering research.”; “A combination of theory and experiments can give insights to the physics of the problem.”; and “As soon as turbulence is involved there will be always an empirical part in the determination of correlations.”

I did not understand yet what your opinion (if any) is for the main question particularly in your area of expertise, Dear Krishnan.

A number of empirical correlations and mechanistic models have been developed for predicting pressure drop and other fluid flow characteristics during multiphase flow in vertical wellbore. At any particular time, engineers are often faced with the difficulty of which particular empirical correlation or mechanistic model should be used to evaluate multiphase flow in vertical wellbore. This is because; a particular correlation or model may be developed under some certain assumptions or for a particular region which may not be applicable to other regions or under a new set of conditions.

https://www.onepetro.org/conference-paper/SPE-136931-MS

Yes empirical methods and correlations are useful, because knowledge can be built on the basis of repeated observations. If we have a large number of observations  over a wide range of circumstances, then we can inference from the empirical (particular) to the theoretical (universal). This is what I said before as Inductive thinking. 

Thanks dear Krishnan for your reference in the field of multiphase flow of oil (published in 2010): “A Comparative Study between Empirical Correlations and Mechanistic Models of Vertical Multiphase Flow.” by Anes Usman Yahaya and Abdallah Al Gahtani. I did not read the paper yet, but the abstract (at least) seems interesting. When I was student, I had an old Indian friend very cult (with oriental culture) very kind and that I liked really very much. Our hopes (and expectative) are sometimes high.

Very interesting debate. In my experience, the development of a fundamental insight on certain problems (e.g. complex types of fluid flow) benefits from developing basic correlations between parameters (preferably non-dimensional) controlling the problem. Some of my past and current experimental work with unsteady multiphase flow has used such correlations. Follow up studies then can use such insights as support in the development of a more formal theoretical framework.

Yes engineering sciences are applied sciences that use theoretical knowledge from fundamental sciences and data  trials to correlate it with practical reality. Usually theories have some degrees of uncertainty, engineering needs fine precision in the development of the processes, products, etc... so some heuristics 'rules of empirical knowledge'  are needed for data adjustments or optimization or fine details.The complexity of the engineering systems and their natural non linearity 'who are usually on a multidisciplinary basis' need fine data correlation for the purpose of their functionality. Heuristics are scientific methods based on knowledge adjustment as well as scientific approaches based on assumptions, theorems and proofs. Heuristics and correlations are as well based on mathematical models

Dear Fairouz: Thank you very much for sharing your practical point of view and a different perspective. I agree that approximations are frequently necessary in common practice (at least in civil and hydraulic engineering) and the degree of such approximations show the adequacy of the formulations.

The empirical methods and correlations are very useful in behavioural data based researches, may not be so useful in engineering, science, etc.

Yes the empirical methods and correlations are used in every branch of science and engineering due to their usefulness and ease.

Yes, it depends on.....

The development of the medical sciences moved from superstition to science practices primarily of classifications to empirical methods based on observation. Prior to the emergence of the medical clinic, diseases were generally conceptualized by Europeans as abstract essences, like Platonic “forms,” that were in some fundamental sense independent of particular bodies. Diseases could become apparent through their visible symptoms, but the visible did not dominate the identification or construction of disease as an object of analysis.

http://www.alcoff.com/content/foucphi.html

Dear Krishnan: Thank you for bringing to this discussion an excerpt and the link of a text written by Linda Martín Alcoff about the Philosophy of Science of Michel Foucault (1926-1984).

http://en.wikipedia.org/wiki/Michel_Foucault

I am sure that a variety of scientific methods to substantiate the necessity of adequate application. Empirical method because of its versatility, as you know, is used in many areas of life. In principle, the life of any person is using the empirical method. The criterion of what I have said is the expression "a person learns from the mistakes." You can only add. Learns only one who admits his mistakes and corrects errors.

Manuel!!! :):)

Dear Manuel: Thanks for your answers and kind words. My sentence you mentioned was relative to a macro scale in the earth context, particularly in the field of Hydraulic Engineering and Civil Engineering. Engineers are normally very pragmatic, and sincerely I have reasons to have a great confidence in the scientific method (as I learned it in the program curricula of philosophy in the secondary school). Of course we can/should always improve our knowledge in every area, as explained by Vladimir, and, in my opinion, I agree that we need to be enough humble for that.

The link below expresses a perspective about “What Science is NOT”:

http://www.indiana.edu/~ensiweb/lessons/unt.not.html

Clocks in computers empirical results demonstrating that real world behaviour diverges from the idealized or assumed behavior

Reliance on computer generated time stamps for correlating events is complicated by uncertainty as to clock skew and drift, factors such as location and local time zone offsets, as well as human factors such as clock tampering. Establishing that a particular computer’s temporal behavior was consistent during its operation remains a challenge. 

https://www.google.co.in/url?sa=t&rct=j&q=&esrc=s&source=web&cd=165&cad=rja&uact=8&ved=0CDwQFjAEOKAB&url=http%3A%2F%2Fwww.dfrws.org%2F2006%2Fproceedings%2F13-%2520Schatz.pdf&ei=n7goVfvWMozfU_-1gMAP&usg=AFQjCNEU4qgo-YbfUkWVS6Xe85WS15QdJw

"For  empirical methods and correlations" - While doing the above you need to assume so many parameters - not all of them is real parameters 

To make the above "useful in the context of the current scientific developments"  - you need to validate it by experimental methods. It is very difficult or sometimes impossible to execute all their assumptions to reality. Almost (except few) never or ever willing to open their mouth about how many empirical parameters or assumptions that  they have introduced while deriving their empirical stuff. 

Actually it depends upon the applications and field. Sometime may end-up in marriage OR may end-up in separation/divorce

Thanks Dr. Murugesan Muthu for your answer with a new perspective of this issue. When working with experimental determination of physical laws (fully scientific based, empirical, or semi empirical) it is in fact very important to select the appropriate variables that are relevant in the laws. A great development in Hydraulics was reached on the basis of experimental research. In general, and whenever possible, relations of cause-effect are clearly highly preferable in comparison to mere empirical correlations between variables.

Dear António, I am not sure of what are you calling "Scientific approach." Do you mean an analytical model? If this is what you mean, from my point of view, the best is using both a model and empirical validation. Unfortunately. as models approach reality, they become closer to intractability, because they are more and more complex, and the required data is much larger in quantity and more difficult to obtain.

Consequently, tractable models are not exact representations of reality, and empirical validation is important to set the estimates of the model parameters.

My two cents.

Cheers,   Vlad