User Haitao Du

58 votes

Regularization methods for logistic regression

answered Aug 8, 2016 at 13:33

54 votes

Why is 600 out of 1000 more convincing than 6 out of 10?

statistical-significance sample-size percentage

answered Jul 28, 2017 at 18:10

43 votes

Accepted

Is there any intuitive explanation of why logistic regression will not work for perfect separation case? And why adding regularization will fix it?

logistic generalized-linear-model optimization intuition separation

answered Oct 13, 2016 at 3:30

38 votes

Accepted

How could stochastic gradient descent save time compared to standard gradient descent?

machine-learning optimization gradient-descent computational-statistics stochastic-gradient-descent

answered Aug 27, 2016 at 15:38

33 votes

Is an overfitted model necessarily useless?

model accuracy overfitting

answered May 11, 2017 at 14:51

28 votes

Are neural networks better than SVMs?

regression machine-learning neural-networks svm

answered Feb 18, 2021 at 10:01

28 votes

Accepted

Boosting: why is the learning rate called a regularization parameter?

machine-learning data-mining predictive-models boosting overfitting

answered Apr 27, 2016 at 20:57

26 votes

Accepted

Do all machine learning algorithms separate data linearly?

machine-learning

answered Jun 1, 2016 at 14:56

26 votes

Accepted

Is low bias in a sample a synonym for high variance?

machine-learning neural-networks variance sampling bias

answered Feb 9, 2018 at 15:15

24 votes

Binary classification with strongly unbalanced classes

machine-learning classification binary-data unbalanced-classes

answered Sep 19, 2016 at 18:55

23 votes

Accepted

How does linear base learner works in boosting? And how does it works in the xgboost library?

regression machine-learning boosting

answered Aug 25, 2016 at 13:52

23 votes

Cohen's kappa in plain English

classification data-mining cohens-kappa

answered Mar 20, 2017 at 15:06

22 votes

Why use gradient descent for linear regression, when a closed-form math solution is available?

regression machine-learning gradient-descent

answered May 10, 2017 at 18:48

22 votes

Accepted

How to decide between PCA and logistic regression?

regression logistic pca

answered Nov 7, 2016 at 15:37

22 votes

Accepted

Why does feature engineering work ?

machine-learning feature-engineering

answered Dec 28, 2017 at 18:01

19 votes

Accepted

what makes neural networks a nonlinear classification model?

neural-networks nonlinear-regression nonlinear

answered Jul 7, 2016 at 18:50

19 votes

What makes a classifier misclassify data?

classification supervised-learning

answered May 23, 2017 at 20:08

15 votes

How do R and Python complement each other in data science?

r python software

answered Oct 6, 2016 at 14:53

14 votes

Accepted

How to know if a learning curve from SVM model suffers from bias or variance?

machine-learning svm bias train

answered Jun 27, 2016 at 13:32

13 votes

What are the four axes on PCA biplot?

r pca biplot

answered Jun 14, 2016 at 14:48

13 votes

To maximize the chance of correctly guessing the result of a coin flip, should I always choose the most probable outcome?

probability

answered May 17, 2016 at 13:53

12 votes

If I want an interpretable model, are there methods other than Linear Regression?

regression machine-learning multiple-regression interpretation

answered Sep 12, 2019 at 5:20

12 votes

How do I know my k-means clustering algorithm is suffering from the curse of dimensionality?

clustering k-means high-dimensional

answered Aug 30, 2016 at 16:55

12 votes

How to run linear regression in a parallel/distributed way for big data setting?

regression linear large-data

answered Feb 22, 2017 at 15:36

12 votes

Why study convex optimization for theoretical machine learning?

machine-learning optimization convex transfer-learning

answered Jan 25, 2018 at 19:40

12 votes

Accepted

Error increase on L2 regularization in an NN

neural-networks regularization training-error

answered Jan 31, 2018 at 17:31

11 votes

What I should do if no distribution fits my dataset?

r distributions fitting

answered Apr 22, 2020 at 16:49

11 votes

Accepted

Can duplicate examples create multi-collinearity?

regression classification linear-model multicollinearity data-preprocessing

answered Oct 27, 2021 at 6:42

11 votes

What does it mean for a linear regression to be statistically significant but has very low r squared?

regression inference r-squared linear

answered Apr 12, 2017 at 20:53

11 votes

Accepted

Avoid overfitting in regression: alternatives to regularization

regression regularization overfitting

answered Jul 19, 2017 at 13:32

Stack Exchange Network

current community

your communities

more stack exchange communities

569 Answers

Regularization methods for logistic regression

Why is 600 out of 1000 more convincing than 6 out of 10?

Is there any intuitive explanation of why logistic regression will not work for perfect separation case? And why adding regularization will fix it?

How could stochastic gradient descent save time compared to standard gradient descent?

Is an overfitted model necessarily useless?

Are neural networks better than SVMs?

Boosting: why is the learning rate called a regularization parameter?

Do all machine learning algorithms separate data linearly?

Is low bias in a sample a synonym for high variance?

Binary classification with strongly unbalanced classes

How does linear base learner works in boosting? And how does it works in the xgboost library?

Cohen's kappa in plain English

Why use gradient descent for linear regression, when a closed-form math solution is available?

How to decide between PCA and logistic regression?

Why does feature engineering work ?

what makes neural networks a nonlinear classification model?

What makes a classifier misclassify data?

How do R and Python complement each other in data science?

How to know if a learning curve from SVM model suffers from bias or variance?

What are the four axes on PCA biplot?

To maximize the chance of correctly guessing the result of a coin flip, should I always choose the most probable outcome?

If I want an interpretable model, are there methods other than Linear Regression?

How do I know my k-means clustering algorithm is suffering from the curse of dimensionality?

How to run linear regression in a parallel/distributed way for big data setting?

Why study convex optimization for theoretical machine learning?

Error increase on L2 regularization in an NN

What I should do if no distribution fits my dataset?

Can duplicate examples create multi-collinearity?

What does it mean for a linear regression to be statistically significant but has very low r squared?

Avoid overfitting in regression: alternatives to regularization