Short-Lived Legendary Lives
When Paris was the new Babylon…
Least squares approach is the backbone of linear regression.
The first time least squares are known to be used is when Legendre created his linear regression model in 1805.
Adrien-Marie Legendre was a mathematician who lived between 1752 and 1833 in France.
Legendre’s life coincides with the years shortly after Thomas Bayes lived and founded Bayes Theorem in Britain. We can see a clear pattern of outstanding developments in statistics lead by French mathematicians in the 18th-19th centuries.
Fun fact: Legendre has a crater on the moon named in his honor: Legendre Crater
His name is also embroiled on the Eiffel Tower along with other French mathematicians such as Laplace, Fourier, Poisson, Lagrange, etc. Now, that’s a way to honor the nation’s geniuses!
Project Gutenberg has the full list of the 72 names along with some interesting criticism. You can check out the full list here.
We see linear regression with least squares first in Niels Henrik Abel’s book: Traité des Fonctions Elliptiques. Abel was a young Norwegian mathematician who took Legendre’s work, studied it, found applications for it, elaborated it and published it. (Sounds familiar? Hint: Bayes-Price)
Check out Naive Bayes History for a similar occasion between Thomas Bayes and Richard Price).
It’s not very surprising to me as I always thought of science and inventions as a big flag race that never ends.
It took thousands and maybe even millions of brilliant minds to arrive to the achievement level we have today and it will keep on happening.
Sadly, Mr. Abel only lived to see his 27 years old age because of tuberculosis that he caught in Paris. Fortunately we have overcome this disease which have claimed so many lives especially in those old days and especially in cities like Paris and London.
("What a beautiful mind has the young Norwegian!")
Linear Regression along with Logistic Regression is definitely one of the oldest and most widespread Machine Learning algorithm out there. Linear Regression has been heavily used in many fields and projects since its discovery in early 1800s.
Another interesting observation I have about the history of Linear Regression is how it’s purely a mathematical (or geometrical specifically) discovery. Legendre was a mathematics and geometry genius and he has made lots of significant contributions directly to the mathematics.
This is slightly different than foundations of other algorithms.
– Bayes also a mathematician, came up with Bayes Theorem while having thoughts about god, chances and existence,
-Verhulst developes Logistic Regression through applications of population growth, laws in sociology and other social sciences,
-Quetelet was also heavily involved and interested in social patterns.
-kNN arises through projects with US military, probably either involved with probability of events, observations etc. or some technnology.
Obviously, these are all noble and breathtakingly impressive work of great humans, but it’s just an interesting observation that Legendre was a more plain vanilla mathematics guy.
I hope you enjoyed this brief article about the history of Linear Regression.
Please feel free to share it on social media (Pinterest, Twitter, Instagram, Quora, Facebook), anything helps! or directly with your friends.
You can also find out about other interesting Machine Learning Algorithms.
Keep educating yourself! It always turns out to be the best investment possible. And great things will happen.