Re: Doubt about results about PCA

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

Re: Doubt about results about PCA

Jovani Souza
Hello,
I have two doubts and would like you to help me please I used PCA in my data and got the following results. 1) How can I interpret these results?


Imagem inline 1
2) How do I interpret the graph of the first principal component vs second component,as shown in the figure below??

Imagem inline 2


Thank you so much guys!

2017-05-26 21:49 GMT-03:00 Jovani T. de Souza <[hidden email]>:
Hello,
I have two doubts and would like you to help me please I used PCA in my data and got the following results. 1) How can I interpret these results?


Imagem inline 1
2) How do I interpret the graph of the first principal component vs second component,as shown in the figure below??

Imagem inline 2


Thank you so much guys!


_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Doubt about results about PCA

Eibe Frank-2
Administrator
1) https://stats.stackexchange.com/questions/31908/what-is-percentage-of-variance-in-pca

   The output you included also shows the coefficients obtained from the eigenvectors, sorted according to absolute value.

2) Your plot shows that the two classes of points cannot be separated accurately based on the first two principal components.

Cheers,
Eibe

> On 27 May 2017, at 13:03, Jovani T. de Souza <[hidden email]> wrote:
>
> Hello,
>
> I have two doubts and would like you to help me please
>
> I used PCA in my data and got the following results.
>
>
> 1) How can I interpret these results?
>
>
>
> <image.png>
> 2) How do I interpret the graph of the first principal component vs second component,as shown in the figure below??
>
> <image.png>
>
>
> Thank you so much guys!
>
> 2017-05-26 21:49 GMT-03:00 Jovani T. de Souza <[hidden email]>:
> Hello,
>
> I have two doubts and would like you to help me please
>
> I used PCA in my data and got the following results.
>
>
> 1) How can I interpret these results?
>
>
>
> <image.png>
> 2) How do I interpret the graph of the first principal component vs second component,as shown in the figure below??
>
> <image.png>
>
>
> Thank you so much guys!
>
> _______________________________________________
> Wekalist mailing list
> Send posts to: [hidden email]
> List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
> List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Doubt about results about PCA

Jovani Souza
Thank you so much Eibe. In relation to the graph, then the data of the first and second components do not present the maximum possible variation? How could visualize better my results in this regard?

Thank you!

2017-05-27 4:48 GMT-03:00 Eibe Frank <[hidden email]>:
1) https://stats.stackexchange.com/questions/31908/what-is-percentage-of-variance-in-pca

   The output you included also shows the coefficients obtained from the eigenvectors, sorted according to absolute value.

2) Your plot shows that the two classes of points cannot be separated accurately based on the first two principal components.

Cheers,
Eibe

> On 27 May 2017, at 13:03, Jovani T. de Souza <[hidden email]> wrote:
>
> Hello,
>
> I have two doubts and would like you to help me please
>
> I used PCA in my data and got the following results.
>
>
> 1) How can I interpret these results?
>
>
>
> <image.png>
> 2) How do I interpret the graph of the first principal component vs second component,as shown in the figure below??
>
> <image.png>
>
>
> Thank you so much guys!
>
> 2017-05-26 21:49 GMT-03:00 Jovani T. de Souza <[hidden email]>:
> Hello,
>
> I have two doubts and would like you to help me please
>
> I used PCA in my data and got the following results.
>
>
> 1) How can I interpret these results?
>
>
>
> <image.png>
> 2) How do I interpret the graph of the first principal component vs second component,as shown in the figure below??
>
> <image.png>
>
>
> Thank you so much guys!
>
> _______________________________________________
> Wekalist mailing list
> Send posts to: [hidden email]
> List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
> List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html


_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Doubt about results about PCA

Eibe Frank-2
Administrator
The x axis in your plot should correspond to the direction with maximum variance in your original data (assuming it shows the first principal component). The y axis should correspond to the (orthogonal) direction with the second-greatest variance.

Cheers,
Eibe

> On 28 May 2017, at 10:38, Jovani T. de Souza <[hidden email]> wrote:
>
> Thank you so much Eibe.
>
> In relation to the graph, then the data of the first and second components do not present the maximum possible variation? How could  visualize better my results in this regard?
>
> Thank you!
>
> 2017-05-27 4:48 GMT-03:00 Eibe Frank <[hidden email]>:
> 1) https://stats.stackexchange.com/questions/31908/what-is-percentage-of-variance-in-pca
>
>    The output you included also shows the coefficients obtained from the eigenvectors, sorted according to absolute value.
>
> 2) Your plot shows that the two classes of points cannot be separated accurately based on the first two principal components.
>
> Cheers,
> Eibe
>
> > On 27 May 2017, at 13:03, Jovani T. de Souza <[hidden email]> wrote:
> >
> > Hello,
> >
> > I have two doubts and would like you to help me please
> >
> > I used PCA in my data and got the following results.
> >
> >
> > 1) How can I interpret these results?
> >
> >
> >
> > <image.png>
> > 2) How do I interpret the graph of the first principal component vs second component,as shown in the figure below??
> >
> > <image.png>
> >
> >
> > Thank you so much guys!
> >
> > 2017-05-26 21:49 GMT-03:00 Jovani T. de Souza <[hidden email]>:
> > Hello,
> >
> > I have two doubts and would like you to help me please
> >
> > I used PCA in my data and got the following results.
> >
> >
> > 1) How can I interpret these results?
> >
> >
> >
> > <image.png>
> > 2) How do I interpret the graph of the first principal component vs second component,as shown in the figure below??
> >
> > <image.png>
> >
> >
> > Thank you so much guys!
> >
> > _______________________________________________
> > Wekalist mailing list
> > Send posts to: [hidden email]
> > List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
> > List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
>
> _______________________________________________
> Wekalist mailing list
> Send posts to: [hidden email]
> List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
> List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
>
> _______________________________________________
> Wekalist mailing list
> Send posts to: [hidden email]
> List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
> List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html