Statistical Analysis Stack Exchange
Statistical Analysis Stack Exchange

Questions tagged [whitening]

9 questions
33
votes
3 answers

Is whitening always good?

A common pre-processing step for machine learning algorithms is whitening of data. It seems like it is always good to do whitening since it de-correlates the data, making it simpler to model. When is whitening not recommended? Note: I'm referring to…
Ran
  • 1,476
  • 3
  • 16
  • 25
4
votes
2 answers

Multiple Regression - Normal equations and features normalization (whitening)

In the multivariate regression analysis, it is easy and natural to conclude that the coefficients of the regression are given by the so-called normal equation $\hat{\beta}=(X^TX)^{-1}X^T y$ My doubt is related to the role of the term $(X^TX)^{-1}$.…
Humberto Fioravante Ferro
  • 151
  • 5
4
votes
0 answers

Is pre-whitening needed for regression on time series data

I am trying to build a regression model for two time series x (independent) and y (dependent) to get y=f(x). I read from other posts saying you have to pre-whitening your time series for both x and y and then make regression on the stationary data…
Vickyyy
  • 416
  • 2
  • 10
2
votes
0 answers

Pre-whitening of AR(2) time series

I'm working on a time-series problem. Consider the model below: $$Y = \beta_0 + \beta_1*t + \beta_2 * \sin(wt) + \beta_3 * \cos(wt) + \epsilon$$ For now, assume that $\epsilon$ is AR(2) with KNOWN parameters $\phi_1$ and $\phi_2$. I want to estimate…
Sam
  • 2,104
  • 19
  • 29
2
votes
0 answers

How to correctly pre-whiten time series

I'm trying to find cross -correlation between two-time series and as it so happens, they are auto-correlated(2), nonstationary and co-integrated. As I read about them, it appears that pre-whitening the series before finding the correlation should be…
harshit
  • 141
  • 3
1
vote
0 answers

Add secondary time series to SARIMA model

I have my primary time series which is monthly temperature. I have the following model for it. fit <- arima(Test, c(1,0,0), seasonal = list(order=c(0,1,1), period =12), include.mean = FALSE) I also have monthly CO2 concentrations over the same…
fighting_fish
  • 11
  • 3
0
votes
0 answers

How do decorrelation stages in data compression work?

I'm trying to understand how decorrelation stages in data compressors work. Many data compression algorithms use decorrelation stages to perform a whitening of the signal before coding it, but WGN is not compressible. Then, my questions are: Why…
Aniol Martí
  • 1
0
votes
0 answers

Data whitening removes linear correlations, but how about non linear one?

Data whitening makes covariance matrix to identity matrix, so linear relationship will be removed through data whitening. This makes learning faster. But how can we remove non linear relationship? Is it enough removing only linear relationship ?
firia2000
  • 131
  • 4
0
votes
0 answers

Pre-whitening longitudinal MLM

I am conducting longitudinal multilevel models (Level-1 = Daily observations; Level-2 = PARTICIPANT). Predictors and outcome variables are measured every day. Do I need to pre-whiten my data (and why)?
George
  • 21
  • 3