From rvosylis at live.com  Wed Jun 10 12:30:38 2015
From: rvosylis at live.com (Rimantas Vosylis)
Date: Wed, 10 Jun 2015 13:30:38 +0300
Subject: [Traminer-users] selecting the number of clusters
Message-ID: <DUB402-EAS39382760B234A62BE260703B4BD0@phx.gbl>

Dear Traminer users,

 

I am trying to build a typology of sequences by using cluster analysis with
OM and Ward algorith.

 

I have a problem of choosing the number of clusters. I use several empirical
indexes, but they don't help me a lot. I use Calinski and harabasz (CH)
index, but it has a peak at two cluster solution and the goes down. I also
use average shilloute width but it gives me the similar results as CH index.
I also run pseudo ANOVA to see which cluster solution explains most
variance, but it tells me the opposite - the more the clusters the higher
the pseudo R2 gets. When I look at the various plots (e.g. seqdplot) I see
that the most meaningful solutions (I have several types of sequences) lie
somewhere between 4-6 clusters.

 

Could You perhaps suggest which indexes worked best for You and matched Your
expectations / theoretical knowledge and that I could use in my analysis?

 

Thank You in advance!!

 

 

Sincerely,

 

Rimantas Vosylis

PhD student, lecturer

Insitute of Psychology

Faculty of Social Technologies

Mykolas Romeris University

 

e-mail: rimantasv at mruni.eu <mailto:rimantasv at mruni.eu> 

e-mail2: rvosylis at live.com <mailto:rvosylis at live.com> 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/c03c9e9e/attachment.html>

From Gerhard.Wuehrer at jku.at  Wed Jun 10 12:55:06 2015
From: Gerhard.Wuehrer at jku.at (=?UTF-8?Q?Gerhard=20W=C3=BChrer?=)
Date: Wed, 10 Jun 2015 12:55:06 +0200
Subject: [Traminer-users] Antw:  selecting the number of clusters
In-Reply-To: <DUB402-EAS39382760B234A62BE260703B4BD0@phx.gbl>
References: <DUB402-EAS39382760B234A62BE260703B4BD0@phx.gbl>
Message-ID: <557833AA02000083000B8488@gwia.im.jku.at>

** Proprietary **
** Reply Requested When Convenient **

Hello,

please try the R - package 'nbclust' do decide how many clusters are
feasable. In addition to that statistical measures, inspect the
different cluster solutions by content and how meaningful
interpretations are. You can also do some kind of x-square test where
you align the clusters with variables not used in the cluster analysis. 
At least you have some kind of face validity. 

Best regards - Gerhard A. W?hrer



o. Univ.-Prof. Dkfm. Dr. Gerhard A. W?hrer
Institut f?r Handel, Absatz und Marketing
Johannes Kepler Universit?t Linz
Altenberger Str. 69
4040 Linz/Austria
tel.: 004373224689401
fax.:004373224689404
mail: gerhard.wuehrer at jku.at
URL: www.marketing.jku.at
 


>>> Rimantas Vosylis <rvosylis at live.com> 10.06.2015 12:30 >>>
Dear Traminer users,
 
I am trying to build a typology of sequences by using cluster analysis
with OM and Ward algorith.
 
I have a problem of choosing the number of clusters. I use several
empirical indexes, but they don?t help me a lot. I use Calinski and
harabasz (CH) index, but it has a peak at two cluster solution and the
goes down. I also use average shilloute width but it gives me the
similar results as CH index. I also run pseudo ANOVA to see which
cluster solution explains most variance, but it tells me the opposite ?
the more the clusters the higher the pseudo R2 gets. When I look at the
various plots (e.g. seqdplot) I see that the most meaningful solutions
(I have several types of sequences) lie somewhere between 4-6 clusters.
 
Could You perhaps suggest which indexes worked best for You and matched
Your expectations / theoretical knowledge and that I could use in my
analysis?
 
Thank You in advance!!
 
 
Sincerely,
 
Rimantas Vosylis
PhD student, lecturer
Insitute of Psychology
Faculty of Social Technologies
Mykolas Romeris University
 
e-mail: rimantasv at mruni.eu
e-mail2: rvosylis at live.com
 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/49e29173/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: W?hrer, Gerhard.vcf
Type: application/octet-stream
Size: 348 bytes
Desc: not available
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/49e29173/attachment.obj>

From rvosylis at live.com  Wed Jun 10 14:36:02 2015
From: rvosylis at live.com (Rimantas Vosylis)
Date: Wed, 10 Jun 2015 15:36:02 +0300
Subject: [Traminer-users] Antw:  selecting the number of clusters
In-Reply-To: <557833AA02000083000B8488@gwia.im.jku.at>
References: <DUB402-EAS39382760B234A62BE260703B4BD0@phx.gbl>
 <557833AA02000083000B8488@gwia.im.jku.at>
Message-ID: <DUB402-EAS36122F97F6746D4B8D4FD5B4BD0@phx.gbl>

Dear Gerhard,

Thank You for this suggestion!

Sincerely

Rimantas

 

From: traminer-users-bounces at lists.r-forge.r-project.org [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of Gerhard W?hrer
Sent: Wednesday, June 10, 2015 1:55 PM
To: traminer-users at lists.r-forge.r-project.org
Subject: [Traminer-users] Antw: selecting the number of clusters

 

Hello,

 

please try the R - package 'nbclust' do decide how many clusters are feasable. In addition to that statistical measures, inspect the different cluster solutions by content and how meaningful interpretations are. You can also do some kind of x-square test where you align the clusters with variables not used in the cluster analysis.  At least you have some kind of face validity. 

 

Best regards - Gerhard A. W?hrer

 

 





o. Univ.-Prof. Dkfm. Dr. Gerhard A. W?hrer
Institut f?r Handel, Absatz und Marketing
Johannes Kepler Universit?t Linz
Altenberger Str. 69
4040 Linz/Austria
tel.: 004373224689401
fax.:004373224689404
mail: gerhard.wuehrer at jku.at <mailto:gerhard.wuehrer at jku.at> 

URL: www.marketing.jku.at <http://www.marketing.jku.at> 

 

 

>>> Rimantas Vosylis <rvosylis at live.com <mailto:rvosylis at live.com> > 10.06.2015 12:30 >>>

Dear Traminer users,

 

I am trying to build a typology of sequences by using cluster analysis with OM and Ward algorith.

 

I have a problem of choosing the number of clusters. I use several empirical indexes, but they don?t help me a lot. I use Calinski and harabasz (CH) index, but it has a peak at two cluster solution and the goes down. I also use average shilloute width but it gives me the similar results as CH index. I also run pseudo ANOVA to see which cluster solution explains most variance, but it tells me the opposite ? the more the clusters the higher the pseudo R2 gets. When I look at the various plots (e.g. seqdplot) I see that the most meaningful solutions (I have several types of sequences) lie somewhere between 4-6 clusters.

 

Could You perhaps suggest which indexes worked best for You and matched Your expectations / theoretical knowledge and that I could use in my analysis?

 

Thank You in advance!!

 

 

Sincerely,

 

Rimantas Vosylis

PhD student, lecturer

Insitute of Psychology

Faculty of Social Technologies

Mykolas Romeris University

 

e-mail: rimantasv at mruni.eu <mailto:rimantasv at mruni.eu> 

e-mail2: rvosylis at live.com <mailto:rvosylis at live.com> 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/6c1278b2/attachment-0001.html>

From rvosylis at live.com  Wed Jun 10 16:44:51 2015
From: rvosylis at live.com (Rimantas Vosylis)
Date: Wed, 10 Jun 2015 17:44:51 +0300
Subject: [Traminer-users] Antw:  selecting the number of clusters
In-Reply-To: <DUB402-EAS36122F97F6746D4B8D4FD5B4BD0@phx.gbl>
References: <DUB402-EAS39382760B234A62BE260703B4BD0@phx.gbl>
 <557833AA02000083000B8488@gwia.im.jku.at>
 <DUB402-EAS36122F97F6746D4B8D4FD5B4BD0@phx.gbl>
Message-ID: <DUB402-EAS250C8AE4E480813EED7021CB4BD0@phx.gbl>

Dear Professor Gerhard,

I tried the NbClust package, but it does not seem to work for analysis of the sequences. Thing is that it has one mandatory argument data which is used to indicate the dataset. However, in sequence analysis this is the sequences of numbers/symbols rather than the vector(s) of numeric variable values. Even though it is possible to specify the distance matrix, it still requires the actual dataset and in my impression, it is not possible to overcome this. 

If you have successfully used this package for sequence analysis, could You possibly copy paste the function that You have used for the calculation of the fit indices?

Thank You in advance!

Rimantas

From: traminer-users-bounces at lists.r-forge.r-project.org [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of Rimantas Vosylis
Sent: Wednesday, June 10, 2015 3:36 PM
To: 'Users questions'
Subject: Re: [Traminer-users] Antw: selecting the number of clusters

 

Dear Gerhard,

Thank You for this suggestion!

Sincerely

Rimantas

 

From: traminer-users-bounces at lists.r-forge.r-project.org <mailto:traminer-users-bounces at lists.r-forge.r-project.org>  [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of Gerhard W?hrer
Sent: Wednesday, June 10, 2015 1:55 PM
To: traminer-users at lists.r-forge.r-project.org <mailto:traminer-users at lists.r-forge.r-project.org> 
Subject: [Traminer-users] Antw: selecting the number of clusters

 

Hello,

 

please try the R - package 'nbclust' do decide how many clusters are feasable. In addition to that statistical measures, inspect the different cluster solutions by content and how meaningful interpretations are. You can also do some kind of x-square test where you align the clusters with variables not used in the cluster analysis.  At least you have some kind of face validity. 

 

Best regards - Gerhard A. W?hrer

 

 

 

o. Univ.-Prof. Dkfm. Dr. Gerhard A. W?hrer
Institut f?r Handel, Absatz und Marketing
Johannes Kepler Universit?t Linz
Altenberger Str. 69
4040 Linz/Austria
tel.: 004373224689401
fax.:004373224689404
mail: gerhard.wuehrer at jku.at <mailto:gerhard.wuehrer at jku.at> 

URL: www.marketing.jku.at <http://www.marketing.jku.at> 

 

 

>>> Rimantas Vosylis <rvosylis at live.com <mailto:rvosylis at live.com> > 10.06.2015 12:30 >>>

Dear Traminer users,

 

I am trying to build a typology of sequences by using cluster analysis with OM and Ward algorith.

 

I have a problem of choosing the number of clusters. I use several empirical indexes, but they don?t help me a lot. I use Calinski and harabasz (CH) index, but it has a peak at two cluster solution and the goes down. I also use average shilloute width but it gives me the similar results as CH index. I also run pseudo ANOVA to see which cluster solution explains most variance, but it tells me the opposite ? the more the clusters the higher the pseudo R2 gets. When I look at the various plots (e.g. seqdplot) I see that the most meaningful solutions (I have several types of sequences) lie somewhere between 4-6 clusters.

 

Could You perhaps suggest which indexes worked best for You and matched Your expectations / theoretical knowledge and that I could use in my analysis?

 

Thank You in advance!!

 

 

Sincerely,

 

Rimantas Vosylis

PhD student, lecturer

Insitute of Psychology

Faculty of Social Technologies

Mykolas Romeris University

 

e-mail: rimantasv at mruni.eu <mailto:rimantasv at mruni.eu> 

e-mail2: rvosylis at live.com <mailto:rvosylis at live.com> 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/510edbd7/attachment.html>

From Gerhard.Wuehrer at jku.at  Wed Jun 10 17:04:13 2015
From: Gerhard.Wuehrer at jku.at (=?UTF-8?Q?Gerhard=20W=C3=BChrer?=)
Date: Wed, 10 Jun 2015 17:04:13 +0200
Subject: [Traminer-users] Antw: Re: Antw: selecting the number of clusters
In-Reply-To: <DUB402-EAS250C8AE4E480813EED7021CB4BD0@phx.gbl>
References: <DUB402-EAS39382760B234A62BE260703B4BD0@phx.gbl>
 <557833AA02000083000B8488@gwia.im.jku.at>
 <DUB402-EAS36122F97F6746D4B8D4FD5B4BD0@phx.gbl>
 <DUB402-EAS250C8AE4E480813EED7021CB4BD0@phx.gbl>
Message-ID: <55786E0D02000083000B84DC@gwia.im.jku.at>

** Proprietary **
** Reply Requested When Convenient **

Dear Rimantas,

I used nbclust for other distance matrices, originating from  other
cluster/segment analysis. If you have the distance matrix, I think you
can input that into nbclust? It may also happen, that there are really
now clusters and the increase of the errors sum follows a monotone
pattern. Please have also a look at the additional literature to be
found with the traminer-package.


Best regards - Gerhard





o. Univ.-Prof. Dkfm. Dr. Gerhard A. W?hrer
Institut f?r Handel, Absatz und Marketing
Johannes Kepler Universit?t Linz
Altenberger Str. 69
4040 Linz/Austria
tel.: 004373224689401
fax.:004373224689404
mail: gerhard.wuehrer at jku.at
URL: www.marketing.jku.at
 


>>> Rimantas Vosylis <rvosylis at live.com> 10.06.2015 16:44 >>>
Dear Professor Gerhard,
I tried the NbClust package, but it does not seem to work for analysis
of the sequences. Thing is that it has one mandatory argument data which
is used to indicate the dataset. However, in sequence analysis this is
the sequences of numbers/symbols rather than the vector(s) of numeric
variable values. Even though it is possible to specify the distance
matrix, it still requires the actual dataset and in my impression, it is
not possible to overcome this. 
If you have successfully used this package for sequence analysis, could
You possibly copy paste the function that You have used for the
calculation of the fit indices?
Thank You in advance!
Rimantas
From: traminer-users-bounces at lists.r-forge.r-project.org
[mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of
Rimantas Vosylis
Sent: Wednesday, June 10, 2015 3:36 PM
To: 'Users questions'
Subject: Re: [Traminer-users] Antw: selecting the number of clusters
 
Dear Gerhard,
Thank You for this suggestion!
Sincerely
Rimantas
 
From: traminer-users-bounces at lists.r-forge.r-project.org
[mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of
Gerhard W?hrer
Sent: Wednesday, June 10, 2015 1:55 PM
To: traminer-users at lists.r-forge.r-project.org
Subject: [Traminer-users] Antw: selecting the number of clusters
 
Hello,
 
please try the R - package 'nbclust' do decide how many clusters are
feasable. In addition to that statistical measures, inspect the
different cluster solutions by content and how meaningful
interpretations are. You can also do some kind of x-square test where
you align the clusters with variables not used in the cluster analysis. 
At least you have some kind of face validity. 
 
Best regards - Gerhard A. W?hrer
 
 
 
o. Univ.-Prof. Dkfm. Dr. Gerhard A. W?hrer
Institut f?r Handel, Absatz und Marketing
Johannes Kepler Universit?t Linz
Altenberger Str. 69
4040 Linz/Austria
tel.: 004373224689401
fax.:004373224689404
mail: gerhard.wuehrer at jku.at
URL: www.marketing.jku.at
 
 
>>> Rimantas Vosylis <rvosylis at live.com> 10.06.2015 12:30 >>>
Dear Traminer users,
 
I am trying to build a typology of sequences by using cluster analysis
with OM and Ward algorith.
 
I have a problem of choosing the number of clusters. I use several
empirical indexes, but they don?t help me a lot. I use Calinski and
harabasz (CH) index, but it has a peak at two cluster solution and the
goes down. I also use average shilloute width but it gives me the
similar results as CH index. I also run pseudo ANOVA to see which
cluster solution explains most variance, but it tells me the opposite ?
the more the clusters the higher the pseudo R2 gets. When I look at the
various plots (e.g. seqdplot) I see that the most meaningful solutions
(I have several types of sequences) lie somewhere between 4-6 clusters.
 
Could You perhaps suggest which indexes worked best for You and matched
Your expectations / theoretical knowledge and that I could use in my
analysis?
 
Thank You in advance!!
 
 
Sincerely,
 
Rimantas Vosylis
PhD student, lecturer
Insitute of Psychology
Faculty of Social Technologies
Mykolas Romeris University
 
e-mail: rimantasv at mruni.eu
e-mail2: rvosylis at live.com
 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/a0c640cd/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: W?hrer, Gerhard.vcf
Type: application/octet-stream
Size: 348 bytes
Desc: not available
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/a0c640cd/attachment-0001.obj>

From rvosylis at live.com  Wed Jun 10 17:39:58 2015
From: rvosylis at live.com (Rimantas Vosylis)
Date: Wed, 10 Jun 2015 18:39:58 +0300
Subject: [Traminer-users] Antw: Re: Antw: selecting the number of
	clusters
In-Reply-To: <55786E0D02000083000B84DC@gwia.im.jku.at>
References: <DUB402-EAS39382760B234A62BE260703B4BD0@phx.gbl>
 <557833AA02000083000B8488@gwia.im.jku.at>
 <DUB402-EAS36122F97F6746D4B8D4FD5B4BD0@phx.gbl>
 <DUB402-EAS250C8AE4E480813EED7021CB4BD0@phx.gbl>
 <55786E0D02000083000B84DC@gwia.im.jku.at>
Message-ID: <DUB402-EAS2228B5BFB8A4FAFA24ED23DB4BD0@phx.gbl>

Dear Gerhard,

Indeed it is possible to input the distance martix, but it is also mandatory to specify the data :( I tried to input the sequence object as data but it does not work :(

I will look through the literature You suggested!

Rimantas

 

From: traminer-users-bounces at lists.r-forge.r-project.org [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of Gerhard W?hrer
Sent: Wednesday, June 10, 2015 6:04 PM
To: traminer-users at lists.r-forge.r-project.org
Subject: [Traminer-users] Antw: Re: Antw: selecting the number of clusters

 

Dear Rimantas,

 

I used nbclust for other distance matrices, originating from  other cluster/segment analysis. If you have the distance matrix, I think you can input that into nbclust? It may also happen, that there are really now clusters and the increase of the errors sum follows a monotone pattern. Please have also a look at the additional literature to be found with the traminer-package.

 

 

Best regards - Gerhard

 

 

 

 





o. Univ.-Prof. Dkfm. Dr. Gerhard A. W?hrer
Institut f?r Handel, Absatz und Marketing
Johannes Kepler Universit?t Linz
Altenberger Str. 69
4040 Linz/Austria
tel.: 004373224689401
fax.:004373224689404
mail: gerhard.wuehrer at jku.at <mailto:gerhard.wuehrer at jku.at> 

URL: www.marketing.jku.at <http://www.marketing.jku.at> 

 

 

>>> Rimantas Vosylis <rvosylis at live.com <mailto:rvosylis at live.com> > 10.06.2015 16:44 >>>

Dear Professor Gerhard,

I tried the NbClust package, but it does not seem to work for analysis of the sequences. Thing is that it has one mandatory argument data which is used to indicate the dataset. However, in sequence analysis this is the sequences of numbers/symbols rather than the vector(s) of numeric variable values. Even though it is possible to specify the distance matrix, it still requires the actual dataset and in my impression, it is not possible to overcome this. 

If you have successfully used this package for sequence analysis, could You possibly copy paste the function that You have used for the calculation of the fit indices?

Thank You in advance!

Rimantas

From: traminer-users-bounces at lists.r-forge.r-project.org <mailto:traminer-users-bounces at lists.r-forge.r-project.org>  [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of Rimantas Vosylis
Sent: Wednesday, June 10, 2015 3:36 PM
To: 'Users questions'
Subject: Re: [Traminer-users] Antw: selecting the number of clusters

 

Dear Gerhard,

Thank You for this suggestion!

Sincerely

Rimantas

 

From: traminer-users-bounces at lists.r-forge.r-project.org <mailto:traminer-users-bounces at lists.r-forge.r-project.org>  [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of Gerhard W?hrer
Sent: Wednesday, June 10, 2015 1:55 PM
To: traminer-users at lists.r-forge.r-project.org <mailto:traminer-users at lists.r-forge.r-project.org> 
Subject: [Traminer-users] Antw: selecting the number of clusters

 

Hello,

 

please try the R - package 'nbclust' do decide how many clusters are feasable. In addition to that statistical measures, inspect the different cluster solutions by content and how meaningful interpretations are. You can also do some kind of x-square test where you align the clusters with variables not used in the cluster analysis.  At least you have some kind of face validity. 

 

Best regards - Gerhard A. W?hrer

 

 

 

o. Univ.-Prof. Dkfm. Dr. Gerhard A. W?hrer
Institut f?r Handel, Absatz und Marketing
Johannes Kepler Universit?t Linz
Altenberger Str. 69
4040 Linz/Austria
tel.: 004373224689401
fax.:004373224689404
mail: gerhard.wuehrer at jku.at <mailto:gerhard.wuehrer at jku.at> 

URL: www.marketing.jku.at <http://www.marketing.jku.at> 

 

 

>>> Rimantas Vosylis <rvosylis at live.com <mailto:rvosylis at live.com> > 10.06.2015 12:30 >>>

Dear Traminer users,

 

I am trying to build a typology of sequences by using cluster analysis with OM and Ward algorith.

 

I have a problem of choosing the number of clusters. I use several empirical indexes, but they don?t help me a lot. I use Calinski and harabasz (CH) index, but it has a peak at two cluster solution and the goes down. I also use average shilloute width but it gives me the similar results as CH index. I also run pseudo ANOVA to see which cluster solution explains most variance, but it tells me the opposite ? the more the clusters the higher the pseudo R2 gets. When I look at the various plots (e.g. seqdplot) I see that the most meaningful solutions (I have several types of sequences) lie somewhere between 4-6 clusters.

 

Could You perhaps suggest which indexes worked best for You and matched Your expectations / theoretical knowledge and that I could use in my analysis?

 

Thank You in advance!!

 

 

Sincerely,

 

Rimantas Vosylis

PhD student, lecturer

Insitute of Psychology

Faculty of Social Technologies

Mykolas Romeris University

 

e-mail: rimantasv at mruni.eu <mailto:rimantasv at mruni.eu> 

e-mail2: rvosylis at live.com <mailto:rvosylis at live.com> 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/8e4c39b1/attachment-0001.html>

From anton.perdoncin at gmail.com  Wed Jun 10 18:51:25 2015
From: anton.perdoncin at gmail.com (Anton Perdoncin)
Date: Wed, 10 Jun 2015 18:51:25 +0200
Subject: [Traminer-users] Format of sequences
Message-ID: <55786B0D.1000504@gmail.com>

Hi,

I have sequences in the following format :

ID    BEGIN1        END1            STATE1        BEGIN2        END2    
   STATE2  etc... until 18
1   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960   
   Y    ...
2   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960   
   Y    ...

One line = one individual. Successive episodes = successive columns.

I know that I need to convert dates into numbers: no problem with that.

However, does anyone have any idea on how I could convert such a df into
an STS or SPELL format ?

Thanks!

Best regards,

Anton Perdoncin

From thomas.collas at gmail.com  Wed Jun 10 21:01:27 2015
From: thomas.collas at gmail.com (thomas collas)
Date: Wed, 10 Jun 2015 21:01:27 +0200
Subject: [Traminer-users] Format of sequences
In-Reply-To: <55786B0D.1000504@gmail.com>
References: <55786B0D.1000504@gmail.com>
Message-ID: <CAOXZeXzPs68unOwt3Z-kLmU458kOGNe_yJnLFZCbMy+b7CM4TQ@mail.gmail.com>

Hello Anton,
An easy solution is to build a very short loop (I know R is not made for
loops but it's only 18 iterations) separating each group of  three columns,
turning the headings into common ones (begin/end/state) and pasting each
one below the other.
I hope that helps,
thomas collas

2015-06-10 18:51 GMT+02:00 Anton Perdoncin <anton.perdoncin at gmail.com>:

> Hi,
>
> I have sequences in the following format :
>
> ID    BEGIN1        END1            STATE1        BEGIN2        END2
>    STATE2  etc... until 18
> 1   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
>    Y    ...
> 2   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
>    Y    ...
>
> One line = one individual. Successive episodes = successive columns.
>
> I know that I need to convert dates into numbers: no problem with that.
>
> However, does anyone have any idea on how I could convert such a df into
> an STS or SPELL format ?
>
> Thanks!
>
> Best regards,
>
> Anton Perdoncin
> _______________________________________________
> Traminer-users mailing list
> Traminer-users at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/10327a7a/attachment.html>

From thomas.collas at gmail.com  Wed Jun 10 21:03:05 2015
From: thomas.collas at gmail.com (thomas collas)
Date: Wed, 10 Jun 2015 21:03:05 +0200
Subject: [Traminer-users] Format of sequences
In-Reply-To: <CAOXZeXzPs68unOwt3Z-kLmU458kOGNe_yJnLFZCbMy+b7CM4TQ@mail.gmail.com>
References: <55786B0D.1000504@gmail.com>
 <CAOXZeXzPs68unOwt3Z-kLmU458kOGNe_yJnLFZCbMy+b7CM4TQ@mail.gmail.com>
Message-ID: <CAOXZeXw9RDuE4A0Cqbxx9_Hn-nitxzac9+D1TV0-vUtj+jVcNA@mail.gmail.com>

Addendum : Do not forget to keep the ID column with the three other columns
at each iteration.

2015-06-10 21:01 GMT+02:00 thomas collas <thomas.collas at gmail.com>:

> Hello Anton,
> An easy solution is to build a very short loop (I know R is not made for
> loops but it's only 18 iterations) separating each group of  three columns,
> turning the headings into common ones (begin/end/state) and pasting each
> one below the other.
> I hope that helps,
> thomas collas
>
> 2015-06-10 18:51 GMT+02:00 Anton Perdoncin <anton.perdoncin at gmail.com>:
>
>> Hi,
>>
>> I have sequences in the following format :
>>
>> ID    BEGIN1        END1            STATE1        BEGIN2        END2
>>    STATE2  etc... until 18
>> 1   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
>>    Y    ...
>> 2   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
>>    Y    ...
>>
>> One line = one individual. Successive episodes = successive columns.
>>
>> I know that I need to convert dates into numbers: no problem with that.
>>
>> However, does anyone have any idea on how I could convert such a df into
>> an STS or SPELL format ?
>>
>> Thanks!
>>
>> Best regards,
>>
>> Anton Perdoncin
>> _______________________________________________
>> Traminer-users mailing list
>> Traminer-users at lists.r-forge.r-project.org
>>
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150610/0aacf95c/attachment.html>

From hc at parisgeo.cnrs.fr  Thu Jun 11 08:45:06 2015
From: hc at parisgeo.cnrs.fr (Hadrien Commenges)
Date: Thu, 11 Jun 2015 08:45:06 +0200 (CEST)
Subject: [Traminer-users] Format of sequences
In-Reply-To: <CAOXZeXw9RDuE4A0Cqbxx9_Hn-nitxzac9+D1TV0-vUtj+jVcNA@mail.gmail.com>
References: <55786B0D.1000504@gmail.com>
 <CAOXZeXzPs68unOwt3Z-kLmU458kOGNe_yJnLFZCbMy+b7CM4TQ@mail.gmail.com>
 <CAOXZeXw9RDuE4A0Cqbxx9_Hn-nitxzac9+D1TV0-vUtj+jVcNA@mail.gmail.com>
Message-ID: <812216976.2959586.1434005106503.JavaMail.zimbra@parisgeo.cnrs.fr>

You could also split your table by set of columns (columns 1:4, then col c(1, 5:7), etc.) and then rbind() all the tables. 

Another option would be the melt() function in the reshape2 package. 

----- Mail original -----

De: "thomas collas" <thomas.collas at gmail.com> 
?: "Users questions" <traminer-users at lists.r-forge.r-project.org> 
Envoy?: Mercredi 10 Juin 2015 21:03:05 
Objet: Re: [Traminer-users] Format of sequences 

Addendum : Do not forget to keep the ID column with the three other columns at each iteration. 

2015-06-10 21:01 GMT+02:00 thomas collas < thomas.collas at gmail.com > : 



Hello Anton, 
An easy solution is to build a very short loop (I know R is not made for loops but it's only 18 iterations) separating each group of three columns, turning the headings into common ones (begin/end/state) and pasting each one below the other. 
I hope that helps, 
thomas collas 

2015-06-10 18:51 GMT+02:00 Anton Perdoncin < anton.perdoncin at gmail.com > : 

<blockquote>
Hi, 

I have sequences in the following format : 

ID BEGIN1 END1 STATE1 BEGIN2 END2 
STATE2 etc... until 18 
1 01/01/1950 01/01/1960 X 02/01/1960 30/01/1960 
Y ... 
2 01/01/1950 01/01/1960 X 02/01/1960 30/01/1960 
Y ... 

One line = one individual. Successive episodes = successive columns. 

I know that I need to convert dates into numbers: no problem with that. 

However, does anyone have any idea on how I could convert such a df into 
an STS or SPELL format ? 

Thanks! 

Best regards, 

Anton Perdoncin 
_______________________________________________ 
Traminer-users mailing list 
Traminer-users at lists.r-forge.r-project.org 
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users 





</blockquote>



_______________________________________________ 
Traminer-users mailing list 
Traminer-users at lists.r-forge.r-project.org 
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150611/ba39ca05/attachment.html>

From thomas.collas at gmail.com  Thu Jun 11 08:53:14 2015
From: thomas.collas at gmail.com (thomas collas)
Date: Thu, 11 Jun 2015 08:53:14 +0200
Subject: [Traminer-users] Format of sequences
In-Reply-To: <812216976.2959586.1434005106503.JavaMail.zimbra@parisgeo.cnrs.fr>
References: <55786B0D.1000504@gmail.com>
 <CAOXZeXzPs68unOwt3Z-kLmU458kOGNe_yJnLFZCbMy+b7CM4TQ@mail.gmail.com>
 <CAOXZeXw9RDuE4A0Cqbxx9_Hn-nitxzac9+D1TV0-vUtj+jVcNA@mail.gmail.com>
 <812216976.2959586.1434005106503.JavaMail.zimbra@parisgeo.cnrs.fr>
Message-ID: <CAOXZeXygNJXE-7s+5beBepWZwCPUif0AcTbWVrrRgMgHfTAPqQ@mail.gmail.com>

Hello,
That's the same proposition actually.
tc

2015-06-11 8:45 GMT+02:00 Hadrien Commenges <hc at parisgeo.cnrs.fr>:

> You could also split your table by set of columns (columns 1:4, then col
> c(1, 5:7), etc.) and then rbind() all the tables.
>
> Another option would be the melt() function in the reshape2 package.
>
> ------------------------------
> *De: *"thomas collas" <thomas.collas at gmail.com>
> *?: *"Users questions" <traminer-users at lists.r-forge.r-project.org>
> *Envoy?: *Mercredi 10 Juin 2015 21:03:05
> *Objet: *Re: [Traminer-users] Format of sequences
>
>
> Addendum : Do not forget to keep the ID column with the three other
> columns at each iteration.
>
> 2015-06-10 21:01 GMT+02:00 thomas collas <thomas.collas at gmail.com>:
>
>> Hello Anton,
>> An easy solution is to build a very short loop (I know R is not made for
>> loops but it's only 18 iterations) separating each group of  three columns,
>> turning the headings into common ones (begin/end/state) and pasting each
>> one below the other.
>> I hope that helps,
>> thomas collas
>>
>> 2015-06-10 18:51 GMT+02:00 Anton Perdoncin <anton.perdoncin at gmail.com>:
>>
>>> Hi,
>>>
>>> I have sequences in the following format :
>>>
>>> ID    BEGIN1        END1            STATE1        BEGIN2        END2
>>>    STATE2  etc... until 18
>>> 1   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
>>>    Y    ...
>>> 2   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
>>>    Y    ...
>>>
>>> One line = one individual. Successive episodes = successive columns.
>>>
>>> I know that I need to convert dates into numbers: no problem with that.
>>>
>>> However, does anyone have any idea on how I could convert such a df into
>>> an STS or SPELL format ?
>>>
>>> Thanks!
>>>
>>> Best regards,
>>>
>>> Anton Perdoncin
>>> _______________________________________________
>>> Traminer-users mailing list
>>> Traminer-users at lists.r-forge.r-project.org
>>>
>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>>>
>>
>>
>
> _______________________________________________
> Traminer-users mailing list
> Traminer-users at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>
>
> _______________________________________________
> Traminer-users mailing list
> Traminer-users at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150611/f4e6c83b/attachment-0001.html>

From Gilbert.Ritschard at unige.ch  Thu Jun 11 08:57:12 2015
From: Gilbert.Ritschard at unige.ch (Gilbert Ritschard)
Date: Thu, 11 Jun 2015 06:57:12 +0000
Subject: [Traminer-users] Format of sequences
In-Reply-To: <812216976.2959586.1434005106503.JavaMail.zimbra@parisgeo.cnrs.fr>
References: <55786B0D.1000504@gmail.com>
 <CAOXZeXzPs68unOwt3Z-kLmU458kOGNe_yJnLFZCbMy+b7CM4TQ@mail.gmail.com>
 <CAOXZeXw9RDuE4A0Cqbxx9_Hn-nitxzac9+D1TV0-vUtj+jVcNA@mail.gmail.com>
 <812216976.2959586.1434005106503.JavaMail.zimbra@parisgeo.cnrs.fr>
Message-ID: <66ABD43696E3DB4687E0BB396A76E5F10A1093E0@golf.isis.unige.ch>

You could also consider the HSPELL_to_STS function provided by the TraMineRextras package.


From: traminer-users-bounces at lists.r-forge.r-project.org [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of Hadrien Commenges
Sent: Thursday, June 11, 2015 08:45
To: Users questions
Subject: Re: [Traminer-users] Format of sequences

You could also split your table by set of columns (columns 1:4, then col c(1, 5:7), etc.) and then rbind() all the tables.

Another option would be the melt() function in the reshape2 package.

________________________________
De: "thomas collas" <thomas.collas at gmail.com<mailto:thomas.collas at gmail.com>>
?: "Users questions" <traminer-users at lists.r-forge.r-project.org<mailto:traminer-users at lists.r-forge.r-project.org>>
Envoy?: Mercredi 10 Juin 2015 21:03:05
Objet: Re: [Traminer-users] Format of sequences

Addendum : Do not forget to keep the ID column with the three other columns at each iteration.

2015-06-10 21:01 GMT+02:00 thomas collas <thomas.collas at gmail.com<mailto:thomas.collas at gmail.com>>:

Hello Anton,
An easy solution is to build a very short loop (I know R is not made for loops but it's only 18 iterations) separating each group of  three columns, turning the headings into common ones (begin/end/state) and pasting each one below the other.
I hope that helps,
thomas collas

2015-06-10 18:51 GMT+02:00 Anton Perdoncin <anton.perdoncin at gmail.com<mailto:anton.perdoncin at gmail.com>>:

Hi,

I have sequences in the following format :

ID    BEGIN1        END1            STATE1        BEGIN2        END2
   STATE2  etc... until 18
1   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
   Y    ...
2   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
   Y    ...

One line = one individual. Successive episodes = successive columns.

I know that I need to convert dates into numbers: no problem with that.

However, does anyone have any idea on how I could convert such a df into
an STS or SPELL format ?

Thanks!

Best regards,

Anton Perdoncin
_______________________________________________
Traminer-users mailing list
Traminer-users at lists.r-forge.r-project.org<mailto:Traminer-users at lists.r-forge.r-project.org>
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users



_______________________________________________
Traminer-users mailing list
Traminer-users at lists.r-forge.r-project.org<mailto:Traminer-users at lists.r-forge.r-project.org>
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150611/c52d2d7c/attachment.html>

From Gilbert.Ritschard at unige.ch  Thu Jun 11 09:06:19 2015
From: Gilbert.Ritschard at unige.ch (Gilbert Ritschard)
Date: Thu, 11 Jun 2015 07:06:19 +0000
Subject: [Traminer-users] Antw: Re: Antw: selecting the number
	of	clusters
In-Reply-To: <DUB402-EAS2228B5BFB8A4FAFA24ED23DB4BD0@phx.gbl>
References: <DUB402-EAS39382760B234A62BE260703B4BD0@phx.gbl>
 <557833AA02000083000B8488@gwia.im.jku.at>
 <DUB402-EAS36122F97F6746D4B8D4FD5B4BD0@phx.gbl>
 <DUB402-EAS250C8AE4E480813EED7021CB4BD0@phx.gbl>
 <55786E0D02000083000B84DC@gwia.im.jku.at>
 <DUB402-EAS2228B5BFB8A4FAFA24ED23DB4BD0@phx.gbl>
Message-ID: <66ABD43696E3DB4687E0BB396A76E5F10A109423@golf.isis.unige.ch>

Did you look at the possibilities offered by the WeightedCluster package of Matthias Studer? The package comes with a vignette that nicely documents the proposed tools.
Gilbert

From: traminer-users-bounces at lists.r-forge.r-project.org [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of Rimantas Vosylis
Sent: Wednesday, June 10, 2015 17:40
To: 'Users questions'
Subject: Re: [Traminer-users] Antw: Re: Antw: selecting the number of clusters

Dear Gerhard,
Indeed it is possible to input the distance martix, but it is also mandatory to specify the data ? I tried to input the sequence object as data but it does not work ?
I will look through the literature You suggested!
Rimantas

From: traminer-users-bounces at lists.r-forge.r-project.org<mailto:traminer-users-bounces at lists.r-forge.r-project.org> [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of Gerhard W?hrer
Sent: Wednesday, June 10, 2015 1:55 PM
To: traminer-users at lists.r-forge.r-project.org<mailto:traminer-users at lists.r-forge.r-project.org>
Subject: [Traminer-users] Antw: selecting the number of clusters

Hello,

please try the R - package 'nbclust' do decide how many clusters are feasable. In addition to that statistical measures, inspect the different cluster solutions by content and how meaningful interpretations are. You can also do some kind of x-square test where you align the clusters with variables not used in the cluster analysis.  At least you have some kind of face validity.

Best regards - Gerhard A. W?hrer



o. Univ.-Prof. Dkfm. Dr. Gerhard A. W?hrer
Institut f?r Handel, Absatz und Marketing
Johannes Kepler Universit?t Linz
Altenberger Str. 69
4040 Linz/Austria
tel.: 004373224689401
fax.:004373224689404
mail: gerhard.wuehrer at jku.at<mailto:gerhard.wuehrer at jku.at>
URL: www.marketing.jku.at<http://www.marketing.jku.at>


>>> Rimantas Vosylis <rvosylis at live.com<mailto:rvosylis at live.com>> 10.06.2015 12:30 >>>
Dear Traminer users,

I am trying to build a typology of sequences by using cluster analysis with OM and Ward algorith.

I have a problem of choosing the number of clusters. I use several empirical indexes, but they don?t help me a lot. I use Calinski and harabasz (CH) index, but it has a peak at two cluster solution and the goes down. I also use average shilloute width but it gives me the similar results as CH index. I also run pseudo ANOVA to see which cluster solution explains most variance, but it tells me the opposite ? the more the clusters the higher the pseudo R2 gets. When I look at the various plots (e.g. seqdplot) I see that the most meaningful solutions (I have several types of sequences) lie somewhere between 4-6 clusters.

Could You perhaps suggest which indexes worked best for You and matched Your expectations / theoretical knowledge and that I could use in my analysis?

Thank You in advance!!


Sincerely,

Rimantas Vosylis
PhD student, lecturer
Insitute of Psychology
Faculty of Social Technologies
Mykolas Romeris University

e-mail: rimantasv at mruni.eu<mailto:rimantasv at mruni.eu>
e-mail2: rvosylis at live.com<mailto:rvosylis at live.com>


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150611/4b1f98ff/attachment-0001.html>

From rvosylis at live.com  Thu Jun 11 11:03:07 2015
From: rvosylis at live.com (Rimantas Vosylis)
Date: Thu, 11 Jun 2015 09:03:07 +0000
Subject: [Traminer-users] Antw: Re: Antw: selecting the number of
 clusters
In-Reply-To: <66ABD43696E3DB4687E0BB396A76E5F10A109423@golf.isis.unige.ch>
References: <DUB402-EAS39382760B234A62BE260703B4BD0@phx.gbl>,
 <557833AA02000083000B8488@gwia.im.jku.at>,
 <DUB402-EAS36122F97F6746D4B8D4FD5B4BD0@phx.gbl>,
 <DUB402-EAS250C8AE4E480813EED7021CB4BD0@phx.gbl>,
 <55786E0D02000083000B84DC@gwia.im.jku.at>,
 <DUB402-EAS2228B5BFB8A4FAFA24ED23DB4BD0@phx.gbl>,
 <66ABD43696E3DB4687E0BB396A76E5F10A109423@golf.isis.unige.ch>
Message-ID: <DUB126-W82D88025862882D6D3B74EB4BC0@phx.gbl>

Dear Gilbert,indeed I found the WeightedCluster package last night and it did help me bigtime in my analysis. In fact I rerun everything with PAM clustering and got better results. So thanks for this reference - it I would not have found it myself that would be a big help!
Rimantas

From: Gilbert.Ritschard at unige.ch
To: traminer-users at lists.r-forge.r-project.org
Date: Thu, 11 Jun 2015 07:06:19 +0000
Subject: Re: [Traminer-users] Antw: Re: Antw: selecting the number	of	clusters









Did you look at the possibilities offered by the WeightedCluster package of Matthias Studer? The package comes with a vignette that nicely documents the proposed
 tools.
Gilbert
 


From: traminer-users-bounces at lists.r-forge.r-project.org
 [mailto:traminer-users-bounces at lists.r-forge.r-project.org] On Behalf Of 
Rimantas Vosylis

Sent: Wednesday, June 10, 2015 17:40

To: 'Users questions'

Subject: Re: [Traminer-users] Antw: Re: Antw: selecting the number of clusters


 
Dear Gerhard,
Indeed it is possible to input the distance martix, but it is also mandatory to specify the data
L I tried to input the sequence object as data but it does not work
L
I will look through the literature You suggested!
Rimantas





 




From:
traminer-users-bounces at lists.r-forge.r-project.org [mailto:traminer-users-bounces at lists.r-forge.r-project.org]
On Behalf Of Gerhard W?hrer

Sent: Wednesday, June 10, 2015 1:55 PM

To: traminer-users at lists.r-forge.r-project.org

Subject: [Traminer-users] Antw: selecting the number of clusters




 




Hello,




 




please try the R - package 'nbclust' do decide how many clusters are feasable. In addition to that statistical measures,
 inspect the different cluster solutions by content and how meaningful interpretations are. You can also do some kind of x-square test where you align the clusters with variables not used in the cluster analysis.  At least you have some kind of face validity.





 




Best regards - Gerhard A. W?hrer




 




 



 



o. Univ.-Prof. Dkfm. Dr. Gerhard A. W?hrer

Institut f?r Handel, Absatz und Marketing

Johannes Kepler Universit?t Linz

Altenberger Str. 69

4040 Linz/Austria

tel.: 004373224689401

fax.:004373224689404

mail: gerhard.wuehrer at jku.at




URL:
www.marketing.jku.at




 



 



>>> Rimantas Vosylis <rvosylis at live.com> 10.06.2015
 12:30 >>>





Dear Traminer users,




 




I am trying to build a typology of sequences by using cluster analysis with OM and Ward algorith.




 




I have a problem of choosing the number of clusters. I use several empirical indexes, but they don?t help me a lot. I use Calinski
 and harabasz (CH) index, but it has a peak at two cluster solution and the goes down. I also use average shilloute width but it gives me the similar results as CH index. I also run pseudo ANOVA to see which cluster solution explains most variance, but it tells
 me the opposite ? the more the clusters the higher the pseudo R2 gets. When I look at the various plots (e.g. seqdplot) I see that the most meaningful solutions (I have several types of sequences) lie somewhere between 4-6 clusters.




 




Could You perhaps suggest which indexes worked best for You and matched Your expectations / theoretical knowledge and that I could
 use in my analysis?




 




Thank You in advance!!




 




 




Sincerely,




 




Rimantas Vosylis




PhD student, lecturer




Insitute of Psychology




Faculty of Social Technologies




Mykolas Romeris University




 




e-mail:
rimantasv at mruni.eu




e-mail2:
rvosylis at live.com




 




 














_______________________________________________
Traminer-users mailing list
Traminer-users at lists.r-forge.r-project.org
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150611/70224066/attachment.html>

From anton.perdoncin at gmail.com  Thu Jun 11 12:22:17 2015
From: anton.perdoncin at gmail.com (Anton Perdoncin)
Date: Thu, 11 Jun 2015 12:22:17 +0200
Subject: [Traminer-users] Format of sequences
In-Reply-To: <66ABD43696E3DB4687E0BB396A76E5F10A1093E0@golf.isis.unige.ch>
References: <55786B0D.1000504@gmail.com>
 <CAOXZeXzPs68unOwt3Z-kLmU458kOGNe_yJnLFZCbMy+b7CM4TQ@mail.gmail.com>
 <CAOXZeXw9RDuE4A0Cqbxx9_Hn-nitxzac9+D1TV0-vUtj+jVcNA@mail.gmail.com>
 <812216976.2959586.1434005106503.JavaMail.zimbra@parisgeo.cnrs.fr>
 <66ABD43696E3DB4687E0BB396A76E5F10A1093E0@golf.isis.unige.ch>
Message-ID: <55796159.3030502@gmail.com>

Hello,

Thanks a lot Thomas, Hadrien and Gilbert.

The HSPELL_TO_STS function is very usefull indeed, but it supposes that 
begin and end variables be numeric, which is not my case since I have 
dates. And converting dates into numbers originating from the most 
ancient date of the first episode seems to me much quicker with SPELL 
format.

So here is one way of converting my HSPELL data into SPELL. Very close 
to Thomas' suggestion (might not be the most elegant script... but it 
does the job):

d <- read.csv2("~ ... example.csv")

f <- data.frame(ident=NULL,deb=NULL,fin=NULL,etat=NULL,motif=NULL)

i <- 1
for(i in 1:20){
   e <- d[,names(d) %in% 
c("ident",paste0(c("deb","fin","etat","motif"),as.character(i)))]
   names(e) <- c("ident","deb","fin","etat","motif")
   f <- rbind(f,e)
}

# ordering by identifiers
f <- f[order(f$ident),]

# deleting empty rows
f <- subset(f, deb!="")

# counting the number of episodes by individual
dim <- nrow(f)
f <- cbind(f,rep(1,dim))
colnames(f)[6]<-"nbepis"

for (i in 1:(dim-1))
{
   if (f[i+1, 1] == f[i, 1])
   {
     f[i+1, 6] <- f[i, 6]+1
   }

}

# checking that maximum number of episodes is equal to 20
max(f$nbepis)


For those who would like to see the actual result: see the example 
dataset attached.

Best,

Anton





Le 11/06/2015 08:57, Gilbert Ritschard a ?crit :
>
> You could also consider the HSPELL_to_STS function provided by the 
> TraMineRextras package.
>
> *From:*traminer-users-bounces at lists.r-forge.r-project.org 
> [mailto:traminer-users-bounces at lists.r-forge.r-project.org] *On Behalf 
> Of *Hadrien Commenges
> *Sent:* Thursday, June 11, 2015 08:45
> *To:* Users questions
> *Subject:* Re: [Traminer-users] Format of sequences
>
> You could also split your table by set of columns (columns 1:4, then 
> col c(1, 5:7), etc.) and then rbind() all the tables.
>
> Another option would be the melt() function in the reshape2 package.
>
> ------------------------------------------------------------------------
>
> *De: *"thomas collas" <thomas.collas at gmail.com 
> <mailto:thomas.collas at gmail.com>>
> *?: *"Users questions" <traminer-users at lists.r-forge.r-project.org 
> <mailto:traminer-users at lists.r-forge.r-project.org>>
> *Envoy?: *Mercredi 10 Juin 2015 21:03:05
> *Objet: *Re: [Traminer-users] Format of sequences
>
> Addendum : Do not forget to keep the ID column with the three other 
> columns at each iteration.
>
> 2015-06-10 21:01 GMT+02:00 thomas collas <thomas.collas at gmail.com 
> <mailto:thomas.collas at gmail.com>>:
>
>     Hello Anton,
>
>     An easy solution is to build a very short loop (I know R is not
>     made for loops but it's only 18 iterations) separating each group
>     of  three columns, turning the headings into common ones
>     (begin/end/state) and pasting each one below the other.
>
>     I hope that helps,
>
>     thomas collas
>
>     2015-06-10 18:51 GMT+02:00 Anton Perdoncin
>     <anton.perdoncin at gmail.com <mailto:anton.perdoncin at gmail.com>>:
>
>         Hi,
>
>         I have sequences in the following format :
>
>         ID    BEGIN1        END1 STATE1        BEGIN2        END2
>            STATE2  etc... until 18
>         1   01/01/1950  01/01/1960          X      02/01/1960   30/01/1960
>            Y    ...
>         2   01/01/1950  01/01/1960          X      02/01/1960   30/01/1960
>            Y    ...
>
>         One line = one individual. Successive episodes = successive
>         columns.
>
>         I know that I need to convert dates into numbers: no problem
>         with that.
>
>         However, does anyone have any idea on how I could convert such
>         a df into
>         an STS or SPELL format ?
>
>         Thanks!
>
>         Best regards,
>
>         Anton Perdoncin
>         _______________________________________________
>         Traminer-users mailing list
>         Traminer-users at lists.r-forge.r-project.org
>         <mailto:Traminer-users at lists.r-forge.r-project.org>
>         https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>
>
> _______________________________________________
> Traminer-users mailing list
> Traminer-users at lists.r-forge.r-project.org 
> <mailto:Traminer-users at lists.r-forge.r-project.org>
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>
>
>
> _______________________________________________
> Traminer-users mailing list
> Traminer-users at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150611/da64c47b/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: example.csv
Type: text/csv
Size: 2250 bytes
Desc: not available
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150611/da64c47b/attachment.csv>

From simon.paye at sciencespo.fr  Thu Jun 11 09:46:20 2015
From: simon.paye at sciencespo.fr (Simon PAYE)
Date: Thu, 11 Jun 2015 09:46:20 +0200
Subject: [Traminer-users] Format of sequences
In-Reply-To: <CAOXZeXygNJXE-7s+5beBepWZwCPUif0AcTbWVrrRgMgHfTAPqQ@mail.gmail.com>
References: <55786B0D.1000504@gmail.com>
 <CAOXZeXzPs68unOwt3Z-kLmU458kOGNe_yJnLFZCbMy+b7CM4TQ@mail.gmail.com>
 <CAOXZeXw9RDuE4A0Cqbxx9_Hn-nitxzac9+D1TV0-vUtj+jVcNA@mail.gmail.com>
 <812216976.2959586.1434005106503.JavaMail.zimbra@parisgeo.cnrs.fr>
 <CAOXZeXygNJXE-7s+5beBepWZwCPUif0AcTbWVrrRgMgHfTAPqQ@mail.gmail.com>
Message-ID: <CAPjUOjNbf=6cpw7UgDyyM2B+K-GWKWUnT3zhdFRb8wQFJd3irw@mail.gmail.com>

Hi there,
Anton, I just used the HSPELL_to_STS function and it worked well.
I suggest you to try something like this:

seq_sts <- HSPELL_to_STS(data,
    begin=c("BEGIN1","BEGIN2",...),
    end=c("END1","END2",...),
    status=c("STATE1","STATE2",...))

Best,
Simon

2015-06-11 8:53 GMT+02:00 thomas collas <thomas.collas at gmail.com>:

> Hello,
> That's the same proposition actually.
> tc
>
> 2015-06-11 8:45 GMT+02:00 Hadrien Commenges <hc at parisgeo.cnrs.fr>:
>
>> You could also split your table by set of columns (columns 1:4, then col
>> c(1, 5:7), etc.) and then rbind() all the tables.
>>
>> Another option would be the melt() function in the reshape2 package.
>>
>> ------------------------------
>> *De: *"thomas collas" <thomas.collas at gmail.com>
>> *?: *"Users questions" <traminer-users at lists.r-forge.r-project.org>
>> *Envoy?: *Mercredi 10 Juin 2015 21:03:05
>> *Objet: *Re: [Traminer-users] Format of sequences
>>
>>
>> Addendum : Do not forget to keep the ID column with the three other
>> columns at each iteration.
>>
>> 2015-06-10 21:01 GMT+02:00 thomas collas <thomas.collas at gmail.com>:
>>
>>> Hello Anton,
>>> An easy solution is to build a very short loop (I know R is not made for
>>> loops but it's only 18 iterations) separating each group of  three columns,
>>> turning the headings into common ones (begin/end/state) and pasting each
>>> one below the other.
>>> I hope that helps,
>>> thomas collas
>>>
>>> 2015-06-10 18:51 GMT+02:00 Anton Perdoncin <anton.perdoncin at gmail.com>:
>>>
>>>> Hi,
>>>>
>>>> I have sequences in the following format :
>>>>
>>>> ID    BEGIN1        END1            STATE1        BEGIN2        END2
>>>>    STATE2  etc... until 18
>>>> 1   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
>>>>    Y    ...
>>>> 2   01/01/1950  01/01/1960          X         02/01/1960   30/01/1960
>>>>    Y    ...
>>>>
>>>> One line = one individual. Successive episodes = successive columns.
>>>>
>>>> I know that I need to convert dates into numbers: no problem with that.
>>>>
>>>> However, does anyone have any idea on how I could convert such a df into
>>>> an STS or SPELL format ?
>>>>
>>>> Thanks!
>>>>
>>>> Best regards,
>>>>
>>>> Anton Perdoncin
>>>> _______________________________________________
>>>> Traminer-users mailing list
>>>> Traminer-users at lists.r-forge.r-project.org
>>>>
>>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>>>>
>>>
>>>
>>
>> _______________________________________________
>> Traminer-users mailing list
>> Traminer-users at lists.r-forge.r-project.org
>>
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>>
>>
>> _______________________________________________
>> Traminer-users mailing list
>> Traminer-users at lists.r-forge.r-project.org
>>
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>>
>
>
> _______________________________________________
> Traminer-users mailing list
> Traminer-users at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users
>

-- 
--------------------------------------------------------------------------
Tous les courriers ?lectroniques ?mis depuis la messagerie de Sciences Po 
doivent respecter les conditions d'usage.
Pour les consulter rendez-vous sur :
http://www.sciencespo.fr/ressources-numeriques/fr/content/regles-de-confidentialite
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20150611/2734c966/attachment-0001.html>