Many thanks for the very perfect answer Dr. Paul Yarnold and also the given links where the alpha test had been correctly applied. Your impeccable response is impressively recommended.

Saluti, Salman!

Early training in systems engineering inspired the genesis of the ODA paradigm. Rather than guessing at a solution based on hypothetical parameters, operations research methods seek the optimal solution for any specific problem given the available data. This orientation hints that I don't naturally favor a "n of m" one-size-fits-all perspective. I look forward to hearing from experts who find this approach useful in their applications, in the hopes that it may be instrumental--I am always looking for new ways to get to the pot of gold waiting at the end of the rainbow.

I offer two opposing views regarding your query. I hope that I am not missing the point of your question!

The N-of-1 artificially intelligent diary project encourages individual users to track as many symptoms as they personally consider relevant. The on-line system provides a list of many symptoms from which each user selects those which they wish to track, and provides opportunity for each user to add other symptoms (not on the list) which they (also/alternatively) wish to track. Different users select different types and different numbers of symptoms--and symptoms selected initially may be dropped by the user (e.g., if a primary symptom becomes irrelevant), and/or new symptoms may be added (e.g., secondary symptoms come into play). The point of the device is to attend to the specific needs-in-time of every individual, and to base all statistical analyses (linking symptoms to other events which the user tracks and records, and offering specific routes for symptom resolution) using only the data provided by the individual. Initial testing of this "one-malleable-size-fits-one" perspective found this algorithm improved the well-being and functional capacity of people who used the diary.

• https://odajournal.com/2013/09/19/an-internet-based-intervention-for-fibromyalgia-self-management-initial-design-and-alpha-test
• https://odajournal.com/2013/11/07/ascertaining-an-individual-patients-symptom-dominance-hierarchy-analysis-of-raw-longitudinal-data-induces-simpsons-paradox
• https://odajournal.com/2013/09/20/modeling-individual-reactivity-in-serial-designs-changes-in-weather-and-physical-symptoms-in-fibromyalgia

In contrast, the unit-weighted MultiODA algorithm is used with a sample. In addition (literally) to the model intercept, every variable in the model can receive a weight of one (add the value of the variable), negative one (subtract the value of the variable), or zero (ignore the variable). The algorithm identifies the single-variable model which maximizes classification accuracy for the sample; the two-variable model which maximizes classification accuracy for the sample; the three-variable model..., and so forth. We look for the point at which no (or trivial, negligible) additional gain is possible given the actual data. The point is to identify the most efficient model which attains maximum accuracy for as many people as possible based on the data on hand. This is solved using mathematical programming methods. Plenty of research in this area was reported in the 1970's--born of the challenge to land on the moon, made possible by increasingly powerful computers.

This is an "n of m" approach in which "n" and "m" are free to vary in different applications (but are the same for all observations within an application) in order to explicitly maximize model performance. If the application is used with a specific questionnaire, a sensitivity analysis would be used ensure identification of the best "n"--neither too few, nor too many, but just right. Thus the "n" and "m" would be free to vary across different questionnaires, but would be fixed within questionnaire, rather than allowing different people to select n different items.

• Yarnold, P.R., Soltysik, R.C., Lefevre, F, & Martin, G.J. (1998). Predicting in-hospital mortality of patients receiving cardiopulmonary resuscitation: Unit-weighted MultiODA for binary data. Statistics in Medicine, 17, 2405-2414.

Naturally, all maximum-accuracy methods are subject to cross-generalizability analysis. The fourth axiom of novometric theory states the performance of the model is assessed in reproducibility analysis, not in training analysis.

Many thanks for the very perfect answer Dr. Paul Yarnold and also the given links where the alpha test had been correctly applied. Your impeccable response is impressively recommended.

Looking at your question was the first thing I did when I awoke, I was worried that I missed the point and embarrassed myself (again). Many thanks for your most gracious and humbling comment, Salman.

If I may, on a personal note, I have been fascinated with engineers and construction workers since the age of three, when I began to observe with awe their work in creating the modern world (I wanted to be a fire fighter). My uncle (RIP) Peter was a civil engineer (built hydroelectric dams in Brazil) and an environmental engineer (evaluated new methods of waste collection/recycling in America and Europe).

I note with respect and gratitude your work on fire safety--all aspects of fire-fighting is a beloved tradition in my home city, Chicago, which has suffered several horrible fires. Chicagoans are grieved by our neighbors who are losing everything to record flooding (another focus of your work) along the Mississippi River. My Chinese colleagues told me that fire, flood and thieves are traditionally their great fears--as these can take away all of one's possessions. I wish you endless success in your endeavors. If I could be of service, it would be a pleasure and an honor. -- Paul

The answer was exactly what I was looking for before sleeping, and surprised when I woke-up! Once again many thanks for the interesting explanations, comments and the worthy compliments Dr. Paul. More-or-less similar reasoning behind me when I decided to study in the field of Civil-Engineering almost 15 years back (I returned to academic ambient again after an approximately 7-8 years of absence, mostly struggling with new experiences at construction-sites in Iran). It would be reciprocally my great honour too, although there is a long way to become such a professional researcher like you.