least squares
C2Technical / Academic
Definition
Meaning
A statistical method of finding the best-fitting line or curve to a set of data points by minimizing the sum of the squares of the vertical distances (differences) between the observed values and the values provided by the model.
The foundational principle of regression analysis used in mathematics, statistics, economics, and machine learning to estimate relationships between variables and make predictions. It can refer to the method (ordinary least squares, weighted least squares) or the resulting line/parameters.
Linguistics
Semantic Notes
Always used as a compound noun. Refers to a principle, method, or the result of applying that method. Typically pre-modified by adjectives like 'ordinary', 'weighted', 'iteratively reweighted', 'partial', etc.
Dialectal Variation
British vs American Usage
Differences
No significant differences in meaning or usage. Spelling conventions follow standard UK/US patterns for surrounding text (e.g., analyse/analyze).
Connotations
Identical technical connotations in both varieties.
Frequency
Equally high frequency in technical/academic contexts in both regions.
Vocabulary
Collocations
Grammar
Valency Patterns
The [MODEL] was fitted by [METHOD OF] least squares.We performed/ran/carried out a least squares [REGRESSION/ANALYSIS/FIT].The [ESTIMATOR/PARAMETERS] were obtained using least squares.Vocabulary
Synonyms
Strong
Neutral
Weak
Vocabulary
Antonyms
Phrases
Idioms & Phrases
- “Not applicable for this technical term.”
Usage
Context Usage
Business
Used in econometrics for forecasting sales, analysing market trends, or building financial models.
Academic
Core concept in statistics, econometrics, data science, engineering, and physical sciences for data modelling.
Everyday
Virtually never used in everyday conversation.
Technical
The primary context of use. Refers to the algorithm, its implementation, or its results in software like R, Python, SPSS, etc.
Examples
By Part of Speech
verb
British English
- The model parameters were then least-squares fitted to the new data.
- We need to least-squares adjust these observations.
American English
- The model parameters were then least-squares fit to the new data.
- We need to least-squares adjust these observations.
adverb
British English
- The curve was fitted least-squares to minimise error.
- Data points were approximated least-squares.
American English
- The curve was fit least-squares to minimize error.
- Data points were approximated least-squares.
adjective
British English
- We obtained the least-squares estimates for the coefficients.
- A least-squares regression line was plotted.
American English
- We obtained the least-squares estimates for the coefficients.
- A least-squares regression line was plotted.
Examples
By CEFR Level
- Not applicable for A2 level.
- Not applicable for B1 level.
- The scientist used a graph to show the line of best fit, which was found using a method called least squares.
- In our maths class, we learned a simple way to draw a trend line called least squares.
- The econometric model was estimated via ordinary least squares, yielding significant coefficients for price and income.
- To minimise prediction error, the algorithm performs a least squares optimisation at each iteration.
Learning
Memory Aids
Mnemonic
Think of trying to draw the 'least bad' line through scattered dots. You measure each dot's vertical miss, square it (to punish big misses more), and add them all up. The 'least squares' line is the one with the smallest total 'square punishment'.
Conceptual Metaphor
The 'best compromise' or 'line of best fit'. Minimising 'squares' metaphorically represents minimising error or disagreement.
Watch out
Common Pitfalls
Translation Traps (for Russian speakers)
- Avoid a literal translation like 'наименьшие квадраты' without the 'метод' (method). The standard term is 'Метод наименьших квадратов (МНК)'.
- Do not confuse with 'квадрат наименьшего значения' which is nonsensical.
Common Mistakes
- Using as a verb (e.g., 'I will least squares the data' – incorrect).
- Confusing 'least squares' (the method) with 'regression' (the broader modelling framework).
- Incorrect pluralisation ('least square' is wrong).
Practice
Quiz
What is the primary goal of the least squares method?
FAQ
Frequently Asked Questions
No, it is primarily a noun phrase (e.g., 'the method of least squares'). It can be used in compound adjectives ('least-squares fit') or, rarely, in hyphenated verb forms ('to least-squares fit'), but it is not a standard standalone verb.
'Least squares' is the general principle. 'Ordinary least squares' (OLS) is the most common specific application of that principle to linear regression models, where all observations are given equal weight.
Squaring the errors serves two main purposes: 1) It penalises larger errors more severely than smaller ones. 2) It ensures all errors are positive, so they can be summed without cancellation, and it makes the mathematical optimisation (finding the minimum) computationally straightforward using calculus.
Yes. The principle of minimising the sum of squared residuals applies to any model where parameters can be adjusted. We then refer to 'non-linear least squares' fitting, which often requires iterative numerical methods, unlike the direct formula for linear least squares.