Principal Component approach discussed in Stock Watson (2002)

[vikas.t...@gmail.com]

Hi Chad,

Wanted to know if the principal component approach on an unbalanced panel (of same frequency) mentioned in Stock and Watson (2002) paper, could be estimated using statsmodels, where missing values are estimated using common component.

[Chad Fulton]

Hello,

The sm.PCA class (https://www.statsmodels.org/dev/generated/statsmodels.multivariate.pca.PCA.html) has an option `missing="fill-em"` that will fill in missing values using the EM algorithm.  This is also what Stock and Watson do, but I am not sure if all of the details are exactly the same.

Best,

Chad

On Tue, Jan 11, 2022 at 3:18 PM vikas.t...@gmail.com <vikas.t...@gmail.com> wrote:

Hi Chad,

Wanted to know if the principal component approach on an unbalanced panel (of same frequency) mentioned in Stock and Watson (2002) paper, could be estimated using statsmodels, where missing values are estimated using common component.

--
You received this message because you are subscribed to the Google Groups "pystatsmodels" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pystatsmodel...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pystatsmodels/9834edfe-66e1-4af3-bdd8-2cd08895c388n%40googlegroups.com.