12

I am describing a new experimental method in a research paper, which has the benefit that all of the input parameters are not arbitrary and are instead determined using basic principles from the available data. As in my subject heading, is there a good word for this? A sample sentence would be "We propose a new ________ method that does not suffer from the arbitrariness of parameter selection that is common in existing approaches."

3
  • What you mean by 'Arbitrariness' Are your competing methodologies' parameter selections based on 'chance', 'whim', or 'impulse.' Are you using it to mean something else? Does the term have a specific meaning in your field which differs from what I understand. I feel like a better understanding of what you are contrasting will help us to find the perfect word for your more reasoned approach. Sep 10 '21 at 18:52
  • I expect the exact word you're looking for is too specific and technical to exist. What seems more suitable to your specific example would be to say something which more generally describes the quality of the method is now better, rather than saying something is now true which wasn't true before (e.g. "We propose a new, more rigorous method...", although I'm not sure whether "rigorous" is the best option here).
    – NotThatGuy
    Sep 12 '21 at 13:19
  • I don't have the rep to answer, but I might use the word systematic in a situation like this, emphasizing that you are not making decisions as you go. I am also reminded of the situation in Machine Learning, where algorithms are described as supervised and unsupervised, according to whether human input is required to make them work. Dec 30 '21 at 6:47

13 Answers 13

30

We propose a new well-founded method that does not suffer from the arbitrariness of parameter selection that is common in existing approaches.

well-founded (adj.)

Based on excellent reasoning, information, judgment, or grounds m-w

If you say that a report, opinion, or feeling is well-founded, you mean that it is based on facts and can therefore be justified. Collins

Having a foundation in fact; based on good reasons, information, etc. dictionary.com


Ecotoxicological studies, especially those concerned with biological monitoring and standardisation of..., are exceedingly difficult in the absence of a uniform, well-founded experimental procedure and quantitative treatment of the results obtained. S. A. Patin; Pollution and the Biological Resources of the Oceans

However, to the best of our knowledge no systematic ways, i.e. based on a well-founded methodology and guiding principles, for building geospatial ontologies have been proposed so far. C. Claramunt et al; GeoSpatial Semantics: 4th International Conference

And a poorly designed experiment leads to a weak scientific result—even the most sophisticated method of analysis cannot change that. On the other hand, a particularly well-qualified cook can still get "that certain something" out of a good recipe and a well-founded analytical method can derive an even more significant scientific insight from a well-design experiment. J. Weimann and J. Brosig-Koch; Methods in Experimental Economics

4
  • Has Collins always phrased their definitions with this colloquial "If you say that..." structure? It's the first time I encounter this in a dictionary.
    – Stef
    Sep 10 '21 at 12:02
  • 3
    I believe their Cobuild dictionary uses this style for what other dictionaries call "usage notes." These are aimed more at learners. This is not the definition style in my print collegiate-size Collins.
    – DjinTonic
    Sep 10 '21 at 12:29
  • 3
    The term "well-founded" has a number of specific (and partly related) meanings within mathematical logic and related fields (including statistics and theory of computation) that have nothing to do with parameter selection. If you're working in those fields, or adjacent to them, I would not use this term unless your method actually matches its specific definition within your field. Sep 10 '21 at 15:35
  • Of course the OP should verify that an adjective is suitable. Since the question contains experimental method, input parameters, and data, I think well-founded is a reasonable candidate, whether or not there is a connection to foundational issues, such as those in math and logic.
    – DjinTonic
    Sep 10 '21 at 15:43
9

rigorous:

OED

3. Extremely detailed and thorough; precise, or concerned with precision; strictly accurate or exact.

1651 T. Hobbes Leviathan i. viii. 34 In Demonstration..and all rigourous Search of Truth, Judgement does all.

1954 H. Becker in H. Becker et al. For Sci. Social Man v. 147 It was so rigorous in method that it still serves as a reprimanding example to those who think that ‘description is easy’.

2003 G. J. Dorrien Making Amer. Liberal Theol. 1900–50 iv. 265 Wieman lectured that in the rigorous sense of the term, all knowledge is scientific.

12
  • 2
    Rigorous does not mean "determined using basic principles from the available data".
    – jimm101
    Sep 10 '21 at 18:22
  • @jimm101 - exactly - that is why it is appropriate.
    – Greybeard
    Sep 10 '21 at 20:07
  • Re-read OP's question?
    – jimm101
    Sep 11 '21 at 14:23
  • 1
    I appreciate your perspective. Reading beyond the title, OP is looking for a term that means "determined using basic principles from the available data". I agree that rigorous is adjacent to a correct answer, and you have provided compelling support for that.
    – jimm101
    Sep 12 '21 at 1:57
  • 2
    I disagree that "rigorous" means "non-arbitrary," or even "sound." The OED definition quoted in this answer doesn't support those meanings.
    – LarsH
    Sep 12 '21 at 10:23
9

The term grounded is well-suited for the purpose,

give (something abstract) a firm theoretical or practical basis

[Google's Oxford scraping (with apologies to the reference experts)]

with a little tweaking of your example sentence. A minor variation would be:

We propose a novel, well grounded method that does not suffer from the arbitrariness of parameter selection that is common in existing approaches.

The term's typical use requires a bit more massaging, but I think it accomplishes your goal.

We propose a novel method that has a well grounded theoretical basis, and does not suffer from the arbitrariness of parameter selection that is common in current approaches.

2
  • 1
    As far as I'm aware, the term "grounded" isn't commonly used in any scientific context like the OP describes. If the OP is looking to coin a neologism, this could be a good option. If they're looking for an established term that readers familiar with their field will (hopefully) understand without explanation, it's not. Sep 10 '21 at 15:38
  • 2
    @IlmariKaronen That's not my experience. DJinTonic's answer of well-founded even smuggles that in: "Based on excellent reasoning, information, judgment, or grounds" [emphasis added]. Certainly you've come across someone basing a proposal on strong theoretical grounds?
    – jimm101
    Sep 10 '21 at 15:44
5

In the physics community, ab initio, Latin for "from the beginning", is widely used to describe research methods built on universal constants rather than arbitrary parameters. It's Latin, but perhaps it's ubiquitous enough to be considered a loanword to English.

For example: ab initio nuclear methods or ab initio quantum methods. A search for "ab initio" on the arXiv, the major preprint server used for physics papers, yields thousands of titles including "ab initio".

You might say

We propose a new ab initio method that does not suffer from the arbitrariness of parameter selection that is common in existing approaches.

2
  • 2
    "ab initio" does not seem to fit what is required. There is nothing in the context to suggest "from first principles". The essence is a method that does not suffer from the arbitrariness of parameter selection
    – Greybeard
    Sep 10 '21 at 11:46
  • The questioner's phrase, italicized in your comment, strongly suggests a method grounded on foundational not arbitrary principles. In fact, several answers posted here rely on this interpretation (e.g. "well-founded", "grounded").
    – mbdeaton
    Sep 10 '21 at 12:04
5

I would say robust:

strong and effective in all or most situations and conditions:

The system requires robust passwords that contain at least one number or symbol. Our goal is to devise robust statistical methods.

(From dictionary.com)1

Your example sentence would become:

  • We propose a new robust method that does not suffer from the arbitrariness of parameter selection that is common in existing approaches.*

This is extremely common phrasing in the literature when talking about statistics, mathematics, experiments, computer systems, etc.

3
  • 6
    The term "robust" is commonly used to mean "resistant to errors", as in "robust statistics" or "robust software" or "robust numerical methods". It does not, in my experience, generally have anything to do with parameter selection (or lack thereof). Sep 10 '21 at 15:27
  • Here I agree with @IlmariKaronen. Although this is extremely common phrasing, it does not mean "determined using basic principles from the available data". To show robustness, you would need an extra step of showing the technique handles a wide range of inputs, not that the initial parameterization isn't arbitrary.
    – jimm101
    Sep 10 '21 at 18:20
  • relevant: en.wikipedia.org/wiki/Robust_statistics eg: a winsorized mean is robust to outliers . a particular estimation method may be robust to departures from normality. i agree with the other comments that this usage seems a bit awkward but it might fit with some rewording -- the estimation method used to estimate the parameters may be more robust than previous methods for example.
    – eps
    Sep 10 '21 at 21:06
3

If you want to emphasize its opposition to "arbitrary," you've already got a great root word in "method." I suggest "methodological," though you wouldn't want to say "methodological method," so maybe change the noun to "approach," "system," etc.

1
  • 2
    Relatedly, I’d suggest methodical as slightly more apt than methodological. The definition you cite notwithstanding, methodological typically refers to something about the methods of a discipline (as the example from dictionary.com illustrates), while methodical means just following such methods.
    – PLL
    Sep 11 '21 at 20:43
3

This will probably depend on the context, but what about parameter-free?

I don't know what your field and the specific context are, but this is the first thing that came to mind when reading your question. Since for your model/method the (former) parameters are now determined from the data, they are fixed for a given dataset and you might not consider then parameters anymore. So you could have a "parameter-free method".

2
  • This seems like the best answer so far to me (with the possible — and possibly biased — exception of my own, which I was writing when this was posted). I'd +2 this if I could. Sep 10 '21 at 15:53
  • or non-parametric?
    – henning
    Sep 11 '21 at 20:18
3

This is really a question about specialized scientific jargon, not about English in general. As such, you should really consult an experienced researched in your field, or review relevant literature to find out what, if any, established term exists for this concept. Any general answers from non-experts are likely to sound wrong or at least unusual to someone familiar with the established terminology in your field of study.

You also haven't specified what your field of research is. From context and phrasing, I'm willing to guess that it's either applied math or statistics, or (probably more likely) some field of experimental or theoretical science that makes heavy use of numerical and statistical tools. But that's about as close as I can get. This makes your question difficult to answer even if one happened to be familiar with it, since it's quite possible for different fields to use different terminology.

That said, since I do happen to have some familiarity with numerical methods and their applications, let me suggest a few possible terms that might work. As it happens, all of these have already been suggested in other answers (some posted while I was typing this), but I figure I should include them for completeness:

  • "Parameter-free", as suggested by Asier R., would be the obvious choice, and is used quite widely for methods like you seem to describe. Of course, this term is strictly appropriate only if your method in fact requires no arbitrary parameters, rather than just fewer of them than usual methods. If it still needs some, I guess you could always call it "quasi-parameter-free"

  • "Ab initio", as suggested by mbdeaton, has a specific established meaning within nuclear physics and related fields such as quantum chemistry. Its meaning is somewhat orthogonal to "parameter-free", and indeed one can find plenty of articles using both terms, but it does involve the same general goal of avoiding the need for arbitrary choices or assumptions by relying exclusively on fundamental physical principles. That said, I would not use this term unless you're working in a field where it is an established term of art, and your method falls within its generally accepted scope in that field.

  • "Rigorous" (suggested by Greybeard) and "theory-based" (suggested by Vaelus) also both have specific connotations that go beyond "requiring no arbitrarily chosen parameters". If your method in fact is rigorous (in the sense of being provably correct and making no unjustified assumptions) or based on theory (rather than empirical results), by all means use them. If not, don't.

However, it's also possible that there is no exact term for the concept you seek, and it's also possible that you don't need one. "We propose a new method that does not suffer from the arbitrariness of parameter selection" is a perfectly fine sentence, after all, even without any adjective between "new" and "method".

1
  • 2
    Saying why a new method doesn't suffer from a common weakness seems preferrable to just saying it doesn't, and the OP is looking for an adjective summing this up.
    – DjinTonic
    Sep 10 '21 at 22:13
2

Theory-based

(adj) theory-based (based in theory rather than experiment)1

"We propose a new theory-based method that does not suffer from the arbitrariness of parameter selection that is common in existing approaches."

2

A method that is "based on sound principles" is principled:

exhibiting, based on, or characterized by principle (Merriam-Webster)

But it must be done in a principled and clear-eyed way. — Annabelle Timsit, Quartz, 23 Aug. 2021

Based on a set of rules. (Wiktionary)

While most of these efforts have focused on compressing the uplink, none of them provide a principled solution to compressing the downlink.

Principled can also have the meaning of "moral, honest," but as a description of a method in a scientific paper, most readers should interpret it as passing judgment only on the soundness of the method.

0

Rigorous
Is a word commonly used in the field of mathematics for describing ideas/theorems etc. which are based on sound principles. Although it may sound contradicting when you consider mathematics is probably the field the with most certain assertions, in the old times, mathematicians sometimes relied on "humanly intuitions" rather than solid mathematical principles. As the field of mathematics has been further improved people started to address this issue and they started to explicitly declare their work as rigorous, which is to say that their work didn't rely upon the so called humanly intuitions.

3
  • Rigorous has already been given as an answer (twice). Sep 10 '21 at 16:18
  • @KillingTime Multiple answers suggesting the same word might still be helpful for the people. For instance, due to the fact that I explain the usage of this word in a distinct context it may people working on specific fields. Sep 10 '21 at 16:25
  • @EfeZaladin The normal pattern here would be to put the additional explanation in a comment under the first. With a little more participation, you'll be able to comment.
    – jimm101
    Sep 13 '21 at 15:28
0

It might sound boastful for you to call your own work this, but an elegant method in science or mathematics is one with “scientific precision, neatness, and simplicity.

This wouldn’t apply if your method is extremely complex, tedious or unintuitive. A phrase that might fit better in that case is, from first principles or derived from first principles.

0

There are a lot of potential words that would fit your description, but for your actual sentence, I'd suggest "data-driven". This fits most nicely within your phrase.

4
  • 2
    I'm not sure that I'd think of "data driven" as meaning "based on sound and not-arbitrary principles". Sep 10 '21 at 14:57
  • You can think that but you'd be wrong.
    – Issel
    Sep 10 '21 at 15:00
  • 1
    I agree, this is the word I was immediately thinking of as well. It would be the right word if the parameters are neither arbitrary nor derived from some theory. At least in physics it is widely used.
    – Graipher
    Sep 11 '21 at 6:02
  • 1
    "Data driven" could mean "determined using basic principles from the available data," but not really "based on sound and not arbitrary principles."
    – LarsH
    Sep 12 '21 at 10:28

Not the answer you're looking for? Browse other questions tagged or ask your own question.