New answers tagged

0 votes

Locally Weighted Linear Regression implementation in either R or Python

Not answering in detail how to implement it, however you can look at the code or just use this library: https://github.com/yaniv-shulman/rsklpr. Note it is quite similar to Lowess but not exactly the ...
yaniv's user avatar
  • 1
1 vote

Calculate decision boundaries and class probabilities based on data

Your outcomes are ordinally scaled. The standard tool is ordinal (logistic) regression. Googling will find you lots of resources to do this in Python, e.g., in statsmodels.
Stephan Kolassa's user avatar
1 vote
Accepted

how can I evaluate the unbalanced data set

There's a bunch of things you could do. Perhaps the simplest is chi-square of the table of the two ratings. There is clearly a strong relationship. The fact that the cells are not even roughly equal ...
Peter Flom's user avatar
  • 110k
0 votes

Performing Word Embeddings with domain-specific data

The study mentioned by PinkBanter has too many flaws in my opinion. Comparing Apples to Oranges: GloVe Wikipedia 100 dim vector, 400K words, GloVe is a completely different method than Word2Vec ...
Jeff Winchell's user avatar
0 votes

using latent dirichlet allocation to reduce the number of dimensions in bag of words model?

EDIT: So after some more research, it appears to me that what they (e.g. the SAM paper) refer to as "topic proportion" is actually alluding the normalized posterior Dirichlet parameters $\...
Kevin's user avatar
  • 51
4 votes

White's test interpretation

Using GARCH does not sound like a great option, because you have not established presence of autoregressive conditional heteroskedasticity (and you have used the ARCH-LM test that could have indicated ...
Richard Hardy's user avatar
7 votes

White's test interpretation

I wouldn't base my choice of model on a test for heteroscedasticity. And I'm not the only one. Here is a quote from the great George Box: To make the preliminary test on variances is rather like ...
Peter Flom's user avatar
  • 110k
2 votes
Accepted

GARCH diagnostics via standardized residuals: interpreting my findings

"As you can see it is not normally distributed." + "I assumed that the residual of the GARCH are t distributed." = What problem do you see in that? I do not see any. For the second ...
Richard Hardy's user avatar
0 votes
Accepted

How come the Bayes Theorem formula results in different probabilities that are verifiable using manual counting?

The first two Python code groups that display either 0.4 or 0.3 are wrong. They do not implement Baye's Theorem. The third group is correct. I was just using Baye's Theorem the wrong way. I wanted to ...
Joachim Rives's user avatar
1 vote
Accepted

standardized residuals GARCH

These are not $p$-values, these are estimated autocorrelations $\hat\rho(h)$ for $h=0,1,\dots,10$. $\hat\rho(0)=\widehat{\text{Corr}}(X_t,X_{t-0})=1$, obviously, thus the value of $1$ at lag $0$. The ...
Richard Hardy's user avatar
2 votes
Accepted

What aspects should I test from a fitted GARCH model?

Residuals vs Standardised Residuals. Residuals are the differences between your observed values and the values predicted by your model. Standardised Residuals in the context of GARCH models are the ...
Robert Long's user avatar
0 votes
Accepted

SARIMAX.predict() and SARIMAX.forecast() exog? Does exog need to be preknown for predict()?

It depends on how you set up your model. To take a very special case of zero-mean ARX(1), it can be $$ y_t=\varphi_1 y_{t-1}+\beta x_t+u_t $$ or $$ y_t=\varphi_1 y_{t-1}+\beta x_{t-1}+v_t. $$ In the ...
Richard Hardy's user avatar
2 votes
Accepted

pdf vs probability vs likelihood

You are absolutly correct that strictly speaking a pdf is not a probability. But this does not matter. All that matters is the pdf is larger where it is "more probable". In other words, ...
user3728501's user avatar

Top 50 recent answers are included