Movatterモバイル変換

Question 1

0votes

0answers

28views

Accuracy increased from including a feature twice? [closed]

I'm using CatBoost. I tested including a feature twice and thrice and it improved accuracy? I tested over many random seeds to confirm it isn't variance. Accuracy by number of times included:0.6675...

blahahaaahahaa

101

askedyesterday

Question 2

4votes

1answer

71views

Is it possible to quanify the minimum amount of information a system must acquire or infer in order to make a reliable prediction about the future?

If a predictive system operates under constraints on information storage, is it possible to formally characterize the minimum sufficient information it must retain or infer from the past in order to ...

Ragul

41

asked2 days ago

Question 3

1vote

0answers

26views

How should I choose a single “best” machine learning model when different models perform best on different OSS projects?

I am evaluating multiple machine learning classifiers on several open-source software (OSS) projects for solving classification problems. For each project, I train and test all models separately and ...

pritpal singh

11

askedDec 13 at 14:45

Question 4

1vote

0answers

13views

Can we assume that training and test loss landscapes look alike?

I am trying to understand why flat minima generalize better than sharp minima, and I have seen figures like the following:I understand that there will be a mismatch between the training loss ...

Antonios Sarikas

951

askedDec 11 at 22:25

Question 5

1vote

0answers

30views

Very small proportions of variance explained by my Multi Factor Analysis (MFA)

I want to analyse the results of a survey in order to explore to what extent do the biases of my population of respondents influence my results, for instance I know that they are younger and more ...

Eliott

11

askedDec 9 at 16:47

Question 6

1vote

1answer

15views

How should distribution shift in docking-derived energy features be handled when ligand size changes?

I’m using docking-derived binding energy values as input features in a machine-learning model.All of the original data was generated from molecules of similar size, but our new dataset contains much ...

CCC

11

askedDec 6 at 21:08

Question 7

1vote

1answer

46views

R Confusion Matrix sensitivity and specificity vs ROC-AUC

I calculated the sensitivity and specificity using the confusionMatrix() function from the caret package, and the results are as follows:...

Math Avengers

657

askedDec 5 at 16:18

Question 8

5votes

1answer

37views

Why are errors in LOOCV more correlated than in k-fold cross-validation?

I’m reading "An Introduction to Statistical Learning" and trying to understand the explanation for why LOOCV typically has higher variance than k-fold cross-validation.ISLR says:When we ...

Antonios Sarikas

951

askedDec 4 at 12:51

Question 9

1vote

0answers

19views

Q2 value in sklearn

I see the Q2 often in the context of PLS, but not really in other areas of machine learning. Nothing in the sklearn docs. Thus my question: If using a 10fold CV scheme on the training set such as ...

dinaue

11

askedDec 3 at 19:49

Question 10

1vote

0answers

18views

Volatility Estimation and forecasting for Ultra-High frequency data

I am unable find methods for Volatility Estimation in ultra-high frequency settings. I am aware HAR-RV and it's counterparts. These models seem to estimate daily volatility using high-frequency data. ...

CuriousRabbit

11

askedDec 3 at 12:31

Question 11

-1votes

1answer

58views

Machine Learning and Hypothesis Testing

In machine learning, do decisions rely on hypothesis tests and p-values?I am uneasy with NHST. The 0.05 threshold is often used by journals without due discretion. To the best of my understanding, ...

Sam

677

askedDec 3 at 11:09

Question 12

8votes

4answers

1kviews

Is Brian Ripley’s Pattern Recognition and NN book still an insightful read?

I have a feeling that the answer is yes, based upon my own preview of it, the way it is cited in more recent works, and of course the many mentions here.I have read other classic texts but still ...

Rick Hass

4,379

askedDec 3 at 0:59

Question 13

0votes

0answers

27views

Difference between Inference, Decision, Estimation, and Learning/Fitting in Generalized Decision Theory?

I am trying to strictly define the relationships between Inference, Decision, Estimation, and Learning/Fitting using the framework of Generalized Bayesian Decision Theory (as taught in MIT 6.437).Set-...

Charlie Parker

7,418

askedDec 2 at 17:40

Question 14

0votes

0answers

10views

Papers to use as references for "acceptable" R² values in temperature and humidity prediction

As the title suggests, I am conducting research using four machine learning models to predict the internal temperature and humidity of an environment. One of my metrics is R², along with RMSE and ...

Iuri Verginio

1

askedDec 2 at 13:34

Question 15

4votes

1answer

198views

What is the difference between the classical decision theoretic setting and the decision theoretic setting in ML and statistical learning?

In classical decision theory one assumes that there is a famility of distributions $\{P_{\theta}: \theta\in\Theta\}$ on a measurable space $(\mathcal{X}, \mathcal{F})$, with $\Theta$ being parameter ...

guest1

1,025

askedDec 2 at 9:22

Movatterモバイル変換

Stack Exchange Network

Questions tagged [machine-learning]

Accuracy increased from including a feature twice? [closed]

Is it possible to quanify the minimum amount of information a system must acquire or infer in order to make a reliable prediction about the future?

How should I choose a single “best” machine learning model when different models perform best on different OSS projects?

Can we assume that training and test loss landscapes look alike?

Very small proportions of variance explained by my Multi Factor Analysis (MFA)

How should distribution shift in docking-derived energy features be handled when ligand size changes?

R Confusion Matrix sensitivity and specificity vs ROC-AUC

Why are errors in LOOCV more correlated than in k-fold cross-validation?

Q2 value in sklearn

Volatility Estimation and forecasting for Ultra-High frequency data

Machine Learning and Hypothesis Testing

Is Brian Ripley’s Pattern Recognition and NN book still an insightful read?

Difference between Inference, Decision, Estimation, and Learning/Fitting in Generalized Decision Theory?

Papers to use as references for "acceptable" R² values in temperature and humidity prediction

What is the difference between the classical decision theoretic setting and the decision theoretic setting in ML and statistical learning?

Hot Network Questions

Subscribe to RSS