How Does Linear Regression Differ Between Statistics and Machine Learning?

AI Thread Summary
The discussion centers on the distinctions between machine learning and statistics, particularly in the context of predictive modeling and interpretability. While both fields utilize similar algorithms, such as linear regression, the approach and emphasis differ. Machine learning typically involves training algorithms on data to enhance predictive accuracy, often utilizing complex models like neural networks that prioritize fitting training data over statistical confidence. In contrast, traditional statistics focuses more on the interpretability of results and the confidence in predictions. The conversation highlights that while linear regression remains fundamentally the same across both domains, machine learning encompasses a broader array of techniques and applications, which may contribute to its perceived superiority and higher marketability compared to traditional statistics.
fog37
Messages
1,566
Reaction score
108
Hello,

I am trying to wrap my head around the difference between machine learning and statistics for predictive purposes and interpretability...

Is there a sharp difference between the two in terms of predictive power? I understand how machine learning needs to be first trained with data to later make useful predictions, etc. Some of the used algorithms are the same we find in statistics (for ex., linear regression).

Given a set of data (not extremely larger), how is linear regression done with machine learning different from the more traditional linear regression?
 
  • Like
Likes WWGD
Technology news on Phys.org
More specifically, how is linear regression done in statistics different from linear regression done with machine learning? Both seem identical processes to predict data we don't have from data we have...
 
Machine learning sells better than statistics.
 
  • Like
Likes phinds
I see...thanks.
 
A lot of machine learning algorithms use neural networks. Neural networks often use functions that are highly nonlinear but allow a better fit to the "training data". There tends to be less emphasis on the statistical confidence of the results. The subject of neural networks is important to understand if you wish to work in machine learning. There is a wide variety of applications, from simply fitting data to having the network develop its own patterns and classifications.
 
  • Like
Likes olgerm
DrStupid said:
Machine learning sells better than statistics.
I disagree with this as a summary of the differences.
 
Last edited:
FactChecker said:
I disagree with this as a summary of the differences.

"How is linear regression done in statistics different from linear regression done with machine learning" in your opinon?
 
  • Like
Likes FactChecker
DrStupid said:
"How is linear regression done in statistics different from linear regression done with machine learning" in your opinon?
Oh, I stand corrected on that part. Linear regression is linear regression. But machine learning and artificial intelligence include so many other techniques that I think it is good to correct the misguided impression of the OP. That may be a reason for a different pay scale.
 

Similar threads

Statistical significance of a ML model...
Replies
2
Views
2K
Python Forecasting Bitcoin (BTC) with Machine Learning
Replies
13
Views
2K
Clustering and Classification in Machine Learning
Replies
29
Views
3K
What are the Limitations of Machine Learning in Causal Analysis?
2
Replies
63
Views
10K
I Linear regression, feature scaling, and regression coefficients
Replies
8
Views
2K
For those who ask: "What programming language should I learn?"
3
Replies
107
Views
8K
Difference between interpreted and compiled files....
Replies
29
Views
3K
I What Are the Key Differences Between OLS and GLM Models in Statistical Analysis?
Replies
3
Views
2K
Linear Regression, etc : Standard vs ML Techniques
Replies
28
Views
3K
Courses Which of these Machine Learning courses should I attend?
Replies
5
Views
2K

Hot Threads

Recent Insights

Back
Top