Questions tagged [robust]

Robustness in general refers to a statistic's insensitivity to deviations from its underlying assumptions (Huber and Ronchetti, 2009).

Filter by
Sorted by
Tagged with
0
votes
0answers
13 views

Are there formal measures for classifier or regression robustness?

Are there performance measures that produce a numerical value of the robustness of a classifier or regression. By robustness I mean graceful degradation in performance to unexpected input (similar to ...
1
vote
2answers
26 views

Robust regression for heavy-tailed random design

As far as I know, there are robust regression methods for outliers in response $Y$ and heavy-tailed error $\epsilon$. The settings for the design matrix (predictor) $X$ is either fixed design or sub-...
2
votes
0answers
13 views

Winsorized mean - trimming furthest points instead of both endpoints

I'm wondering if the Winsorized mean can be improved by trimming the 5% farthest points from the mean instead of trimming 5% on each endpoint. Concretely: Consider the Winsorized mean, where we ...
1
vote
1answer
48 views

Jackknifing for assessing the “robustness” of test results

In a presentation I saw recently, a two-sided t-test was repeated with jackknifed subsets of the original data in order to assess the result's "robustness". In detail, they took a random ...
1
vote
1answer
23 views

Optimization function of the Hodges-Lehmann location estimator

The median minimizes the sum of absolute differences while the mean minimizes the sum of square distances. What is the function which is minimized by the Hodges-Lehmann location estimator?
1
vote
0answers
20 views

Is traditional negative binomial regression robust to model misspecification or not?

By "traditional" NBR I mean NB2, i.e. the one modeling variance as a quadratic function of the mean, with the formula: $Var(Y)=E[Y]*(1+\alpha*E[Y])$. I have found contrasting statements in ...
4
votes
1answer
73 views

Trimmed, weighted mean

The trimmed mean (or truncated mean) is a robust version of the mean, designed to be robust to outliers. I am wondering what is the right trimmed version of a weighted average. If I have a sample ...
0
votes
1answer
15 views

How to robustly present a min and a max value?

I have a set of measurements from an air polution sensor. I want to determine the min and the max value in a period of time (let's say in a day). The min and the max don't have to be the true ...
0
votes
0answers
12 views

Instrumental Variable - clustering and standard errors, in both stages?

I was wondering whether in an instrumental variable procedure, you do the clustering and standard errors in both stages or just the final stage. Wooldridge (fifth edition, section 15.6) in the first ...
5
votes
0answers
63 views

Is there a measure of the robustness of a statistic?

I got a question today when talking about mean and median, IQR and variance. Is there a numerical measure of the robustness of a statistic? I must confess that I had never thought about that before, ...
0
votes
0answers
20 views

When to use robust one-way repeated measure design ANOVA?

I have a set of 9 different factor levels from my independent variable to be compared against each other. Here are the results of the different assumption tests in R. I'm just going through my ...
0
votes
0answers
23 views

Interpretation Robustness Check table

we conducted our robustness check for our negative binomial regression using "checkrob" command in Stata (Barslund, Rand, Tarp; 2007). As a result we got the following table. How can I tell ...
0
votes
0answers
18 views

How to clean up outliers in regression which cannot be visualized?

Recently I meet a problem in an interview. Given a dataset $\{(X_i, y_i) \}$ for regression problem, how to detect and clean up outliers before starting using any regression algorithm. The following ...
0
votes
0answers
26 views

Post hoc test for robust mixed design ANOVA using R

I have calculated a robust mixed ANOVA because I have no homogeneity of the error variances nor of the covariances (with a 2 level between-subject and a 2 level within-subject variable). Can I ...
2
votes
0answers
47 views

In general, how to determine the weight function of Robust regression

I think the question is clear from the title. How the weight function for example in Huber is calculated? Is it by differentiating the objective function?
0
votes
0answers
33 views

Weibull with known shape parameter

I am new to Bayesian robustness. If I have Weibull likelihood $X$~ Weibull($\lambda$, $\beta$)$= \lambda \beta x^{-\beta} \exp(-\lambda x^\beta)$ with $\lambda$ unknown and $\beta$ known. we know that ...
0
votes
0answers
7 views

NA values when converting into wide format for Wilcox's robust ANOVA

I have a dataset of more than 10.000 values, with 12 different factor levels that are NOT evenly distributed. For my dataset, Levene's test is very significant which means that the assumption of ...
0
votes
0answers
10 views

Efficient implementation of spike removal

I have implemented the moving median absolute deviation (moving MAD) and it seems like bit-exact to Matlab's implementation. Nevertheless, I am sure that it is not efficient. The usual median filter ...
3
votes
1answer
228 views

Can we ALWAYS assume normal distribution if n >30?

I'm in a debate with a coworker and I'm starting to wonder if I'm wrong but the internet is confusing me more. We have continuous data [0,infinity) that is retrospectively selected on individuals. The ...
0
votes
0answers
13 views

Weighting observations according to their leverage glmnet

I wish to estimate the association between several biological features and a binary health outcome. These biological features, however, have occasional extreme (but valid) outlying observations. Given ...
0
votes
0answers
11 views

Hodges-lehmann estimator is ist a U statistic?

I want to know if the hodges lehmann estimator is U statistic and if yes how to prove it?
1
vote
1answer
25 views

Is it possible to use `robustlmm` package for other distributions than normal?

I have a response variable that is non-normaly distributed (~Gamma). Due to the fact that I have a lot of "contamination", I would need to use a robust mixed-effects model method that is ...
1
vote
0answers
48 views

How to manually inflate standard errors to approximate clustered SEs

I'm reading a handout on clustering here It's not clear to me how to compute $\rho_x$ or $\rho_\epsilon$. What is meant by within-cluster correlation of the regression, or within-cluster error ...
0
votes
0answers
41 views

Comparing distributions based on specified conditions?

I wish to compare a set of distributions based on two specified conditions. These conditions originate from the theory within the field I am working and can be used to define what is a good/poor ...
2
votes
2answers
60 views

Difference between removing outliers and using Least Trimmed Squares?

In most cases we would be reluctant to remove outliers from the dataset just to get a better fit. Robust estimators such as Least Trimmed Squares are sometimes recommended in order to fit a regression ...
0
votes
1answer
49 views

Data Perturbation - Model robustness test

I came across with presentation about robustness test recently and I didn't exactly understand how to apply it to ML model (not DL). The presentation show a graph: x-axis - some metric, for example ...
0
votes
0answers
7 views

How to find robustness of a matrix with respect to another matrix

I have a matrix $D_{s}$ that has been computed using another matrix $X_{s}$ by solving an optimization problem (in the optimization problem, $X_{s}$ was a data matrix and $D_{s}$ was estimated by the ...
1
vote
0answers
74 views

Loss function vs regularizer

Probably this is an easy thing. But I am struggling with the concept of regularization and loss function. Let's assume that we want a robust sparse solution of a linear regression problem. The data is ...
0
votes
0answers
13 views

Does data augmentation on the incompleted dataset really improves robustness?

Lots of research use data augmentation(DA) during the training phase, and the result shows that expanding the dataset is feasible for model generalization. The goal of DA is to generalize data ...
0
votes
0answers
27 views

Calculation of Olea and Pflueger's (2013) Effective F-statistic

I am trying to understand the calculation of Olea and Pflueger's (2013) test for weak instruments when the errors are not conditionally homoskedastic and serially uncorrelated, which is calculated as: ...
3
votes
1answer
128 views

How are clustered standard errors and Newey-West errors related

As the question says, how are the two concepts related? As far as I understand, both approaches correct for heteroscedasticity and autocorrelation. Yet, they are different. Would applying one of the ...
0
votes
0answers
19 views

Should results from Forward Orthogonal Deviation be close to those using first-differences with lagged dependent variables as instruments?

It seems to me that the Forward Orthogonal Deviation (FOD) transformation is typically compared with the First-Difference (FD) one: https://www.researchgate.net/publication/...
1
vote
0answers
31 views

What is the name for this type of sensitivity measure in regression analysis?

I have a way of calculating sensitivity of a regression that is very useful for my particular domain, but I don't know what it is called. I would like a name for it so I can look up additional ...
1
vote
1answer
68 views

How to choose between different methods of linear regression?

I find following commonly mentioned linear regression methods: OLS: ordinary least squares GLS: generalized least squares WLS: weighted least squaes RLM: robust linear model OLS is usually the default....
0
votes
1answer
144 views

Placebo test for the validity of difference-in-differences analysis results for limited data period

I am doing an analysis of a strategy implemented by a four-year-old corporate company and this implementation was given in different cities which all of its employees were received the treatment or ...
0
votes
0answers
15 views

Which is the best approach for robustness testing for deep learning models?

I tried 10 fold CV and my validation accuracy is above 95%. Just to check the robustness of my CNN model I wanted to do a permutation testing. But is that necessary?
0
votes
0answers
15 views

Scale factor for trimmed standard deviations?

For normally distributed data, what is the scale factor of a trimmed standard deviation? For instance, let's say we have a normal distribution with $\sigma = 1$. If we remove the top and bottom $t<...
7
votes
4answers
374 views

Relationship between overfitting and robustness to outliers

What's the relationship between overfitting and sensitivity to outliers? For example: Does robustness to outliers make necessarily models less prone to overfitting? What about the other way around? ...
1
vote
0answers
54 views

Robust RM ANOVA [closed]

I am an R beginner and I needed to run robust two way mixed ANOVAs with R using the WRS2 package and the function btwin because my data has nonnormal residuals (I looked at Q-Q plots) and there is a ...
1
vote
0answers
42 views

Bootstrapping for rlmer [closed]

I am using the robustlmm package and the rlmer function to calculate robust estimates for my multi-level data (2-levels and 3-...
0
votes
0answers
14 views

Why are maximum likelihood estimation methods more robust to missing data than quasi-likelihood estimation methods?

Why are maximum likelihood estimation methods more robust to missing data than quasi-likelihood estimation methods? In GEE versus Mixed-Effects, Mixed-Effects based on ML is more robust to missing ...
0
votes
1answer
56 views

Which reference could I cite for using confidence intervals instead of P-values when using robust linear mixed models (R package: robustlmm)?

I am using the R package robustlmm for some analysis due to its characteristics. I have read in posts like this one that with this type of models is better to use confidence intervals than $p$-values, ...
2
votes
2answers
74 views

If we are taught to not remove outliers without investigation, how do robust methods (median, trimmed mean) can be even suggested?

I just saw an article, which taught to nor remove outliers without investigation, because it may be a unusual but valid observation or naturally skewed data, for example in chemistry or medicine. Only ...
0
votes
0answers
6 views

Simple GLMMs with imperfect fit---are they robust?

I'm wondering if there is a good and accessible discussion of how robust GLMMs of imperfect fit (e.g. some patterns in residuals) are in simple models, for instance those where repeated measures ANOVA ...
0
votes
0answers
18 views

Robust regression coefficients replication

I am trying to perform a robust regression using the LMS (least median of squares) method. For some reason, I just can't seem to replicate the results the book is showing. The data set the book is ...
1
vote
0answers
17 views

why linear regression gives high mean squared error with robust scaler?

I've been trying to use a pipeline that consists of robust scaler and one of the algorithms like xgboost, ridge, lasso, and linear regression, and they all give better results except for linear ...
1
vote
0answers
13 views

Two-group “ANCOVA” with Theil-Sen?

("ANCOVA" might typically refer to using at least three groups. In what I do, I am always comparing two groups, and that is what I mean throughout this question.) I know how to do ANCOVA via ...
1
vote
0answers
34 views

Measure “concentratedness” of 1D data

I have $N$ datapoint dataset $\{x_i, y_i\}$, where $x_i$ are equally spaced over the interval $[0, 1]$, and $y_i$ are non-negative. It is known that $y_i$ is a sum of a signal and gaussian noise. The ...
0
votes
0answers
40 views

Discrepancies in robust ancova calculated in R with the WRS2 package vs by hand

I've conducted a study where participants in 2 groups (Control and Experimental) completed three scales (NJ, A, and D) at two time points (T1 and T2). I'm running ANCOVAs with the T2 responses as my ...
0
votes
0answers
52 views

Does the sigma-clipped variance/standard-deviation count as a robust estimator of scale?

Robust estimators of scale, such as the median absolute deviation (MAD) and so on, are less affected by outliers than something like the basic standard deviation/variance. Firstly, is there a specific ...

1
2 3 4 5
10