From minal.crri at gmail.com Sun Sep 9 17:19:08 2018
From: minal.crri at gmail.com (minal srivastava)
Date: Sun, 9 Sep 2018 12:49:08 +0530
Subject: [Limdep Nlogit List] Fwd: Joint logit model LIMDEP
In-Reply-To:
References:
Message-ID:
---------- Forwarded message ---------
From: minal srivastava
Date: Fri, Sep 7, 2018, 16:41
Subject: Joint logit model LIMDEP
To:
Dear All,
In case of forming a joint logit model for residential location choice and
travel mode choice, how should my data format look like.
I know for a MNL model or a Nested logit model, for each observation I need
to convert to long data format ( depending upon the number of choice
alternative). For e.g. for 100 observations with 5 travel mode choices, the
total number of data input lines would be (5*100) =500.
But in case of joint logit model, If I have 3 location choice alternatives
and 5 travel mode choice alternative how should I prepare my data input
file? Am I supposed to use two different data files, one with (3*100) 300
input lines and other with (5*100) 500 input lines?
Looking forward to your kind reply.
--
Thanks & regards,
Minal
New Delhi- India.
Email: minal.crri at gmail.com
Website: https://protect-au.mimecast.com/s/RW4LCgZowLHkA9YxTNB5pM?domain=crridom.gov.in
*"**Success is a science; if you have the conditions, you get the result.*"
- *Oscar Wilde*
From minal.crri at gmail.com Sun Sep 9 17:21:20 2018
From: minal.crri at gmail.com (minal srivastava)
Date: Sun, 9 Sep 2018 12:51:20 +0530
Subject: [Limdep Nlogit List] Joint logit model LIMDEP
In-Reply-To:
References:
Message-ID:
---------- Forwarded message ---------
From: minal srivastava
Date: Fri, Sep 7, 2018, 16:41
Subject: Joint logit model LIMDEP
To:
Dear All,
In case of forming a joint logit model for residential location choice and
travel mode choice, how should my data format look like.
I know for a MNL model or a Nested logit model, for each observation I need
to convert to long data format ( depending upon the number of choice
alternative). For e.g. for 100 observations with 5 travel mode choices, the
total number of data input lines would be (5*100) =500.
But in case of joint logit model, If I have 3 location choice alternatives
and 5 travel mode choice alternative how should I prepare my data input
file? Am I supposed to use two different data files, one with (3*100) 300
input lines and other with (5*100) 500 input lines?
Looking forward to your kind reply.
--
Thanks & regards,
Minal
New Delhi- India.
Email: minal.crri at gmail.com
Website: https://protect-au.mimecast.com/s/ht0aC3Q8Z2FBpq9RCg6F97?domain=crridom.gov.in
*"**Success is a science; if you have the conditions, you get the result.*"
- *Oscar Wilde*
From minal.crri at gmail.com Tue Sep 11 04:25:53 2018
From: minal.crri at gmail.com (minal srivastava)
Date: Mon, 10 Sep 2018 23:55:53 +0530
Subject: [Limdep Nlogit List] Joint logit Model in Nlogit
Message-ID: <3BE1FEAD-3100-496F-B0D2-BB6F3015B109@gmail.com>
Dear All,
In case of forming a joint logit model for residential location choice and travel mode choice, how should my data format look like.
I know for a MNL model or a Nested logit model, for each observation I need to convert to long data format ( depending upon the number of choice alternative). For e.g. for 100 observations with 5 travel mode choices, the total number of data input lines would be (5*100) =500.
But in case of joint logit model, If I have 3 location choice alternatives and 5 travel mode choice alternative how should I prepare my data input file? Am I supposed to use two different data files, one with (3*100) 300 input lines and other with (5*100) 500 input lines?
Looking forward to your kind reply.
From minal.crri at gmail.com Tue Sep 11 04:25:53 2018
From: minal.crri at gmail.com (minal srivastava)
Date: Mon, 10 Sep 2018 23:55:53 +0530
Subject: [Limdep Nlogit List] Joint logit Model in Nlogit
Message-ID: <3BE1FEAD-3100-496F-B0D2-BB6F3015B109@gmail.com>
Dear All,
In case of forming a joint logit model for residential location choice and travel mode choice, how should my data format look like.
I know for a MNL model or a Nested logit model, for each observation I need to convert to long data format ( depending upon the number of choice alternative). For e.g. for 100 observations with 5 travel mode choices, the total number of data input lines would be (5*100) =500.
But in case of joint logit model, If I have 3 location choice alternatives and 5 travel mode choice alternative how should I prepare my data input file? Am I supposed to use two different data files, one with (3*100) 300 input lines and other with (5*100) 500 input lines?
Looking forward to your kind reply.
From walibehram at yahoo.com Tue Sep 11 09:27:37 2018
From: walibehram at yahoo.com (Behram Wali)
Date: Mon, 10 Sep 2018 19:27:37 -0400
Subject: [Limdep Nlogit List] Joint logit Model in Nlogit
In-Reply-To: <3BE1FEAD-3100-496F-B0D2-BB6F3015B109@gmail.com>
References: <3BE1FEAD-3100-496F-B0D2-BB6F3015B109@gmail.com>
Message-ID: <971F26CF-18CD-422E-8A90-096D79F3A083@yahoo.com>
Minal, by joint model, do you mean to refer to simultaneous estimation of the two MNL models, one for residential location choice and one for travel mode? If this is what you mean, I don?t think this is pre-coded in NLOGIT. Others may comment. You will have to code the joint model yourself.
One naive way of modeling the joint behavior through a single MNL would be to define the choice sets in a way that considers both of the choice-sets. Say you have two residential location types (conventional and neo-traditional) and three modes (bus, car, shared), you can define distinct choice sets for each combination of the alternatives in two choice-sets. E.g., 1 if location == conventional AND mode == bus, 2 if location == conventional AND mode == shared, 3 if location == neo AND mode == bus, and so on. And then estimate an MNL with these new choice sets as alternatives.
Best, -wali
> On Sep 10, 2018, at 2:25 PM, minal srivastava via Limdep wrote:
>
> Dear All,
>
> In case of forming a joint logit model for residential location choice and travel mode choice, how should my data format look like.
> I know for a MNL model or a Nested logit model, for each observation I need to convert to long data format ( depending upon the number of choice alternative). For e.g. for 100 observations with 5 travel mode choices, the total number of data input lines would be (5*100) =500.
>
> But in case of joint logit model, If I have 3 location choice alternatives and 5 travel mode choice alternative how should I prepare my data input file? Am I supposed to use two different data files, one with (3*100) 300 input lines and other with (5*100) 500 input lines?
>
> Looking forward to your kind reply.
>
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> http://limdep.itls.usyd.edu.au
>
From wgreene at stern.nyu.edu Thu Sep 13 06:06:23 2018
From: wgreene at stern.nyu.edu (William Greene)
Date: Wed, 12 Sep 2018 16:06:23 -0400
Subject: [Limdep Nlogit List] Joint logit Model in Nlogit
In-Reply-To: <3BE1FEAD-3100-496F-B0D2-BB6F3015B109@gmail.com>
References: <3BE1FEAD-3100-496F-B0D2-BB6F3015B109@gmail.com>
Message-ID:
Minal.
It is not established in this thread what is a "joint logit model?" I do
not see the formulation
(equations) of a model that has these two models as components in something
that is
different from two separate MNL models. That has to be nailed down before
we can talk
about how to estimate it. To the point, what is correlated with what that
makes this pair of
settings different from a pair of MNL models. As it stands now, since they
are indeed just
two logit models, the way to set it up is to have the first 500 rows be the
travel mode data,
then rows 501 to 800 be be the data for the location data. Then,
SAMPLE;1-500$ followed by the travel mode analysis, then
SAMPLE; 501-800 $ followed by the residential location modeling.
If you want to follow the suggestion of folding the data into one decision,
you will have to
have a 5x3 = 15 alternative choice model. I'm not yet convinced that this
would be
plausible. Likewise, a nested logit with travel in one branch and location
in the other is also
a nonstarter, since here, every individual would be choosing both branches.
Regards
Bill Greene
On Mon, Sep 10, 2018 at 2:25 PM, minal srivastava via Limdep <
limdep at mailman.sydney.edu.au> wrote:
> Dear All,
>
> In case of forming a joint logit model for residential location choice and
> travel mode choice, how should my data format look like.
> I know for a MNL model or a Nested logit model, for each observation I
> need to convert to long data format ( depending upon the number of choice
> alternative). For e.g. for 100 observations with 5 travel mode choices, the
> total number of data input lines would be (5*100) =500.
>
> But in case of joint logit model, If I have 3 location choice alternatives
> and 5 travel mode choice alternative how should I prepare my data input
> file? Am I supposed to use two different data files, one with (3*100) 300
> input lines and other with (5*100) 500 input lines?
>
> Looking forward to your kind reply.
>
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> http://limdep.itls.usyd.edu.au
>
>
--
William Greene
Department of Economics
Stern School of Business, New York University
44 West 4 St., 7-90
New York, NY, 10012
URL: https://protect-au.mimecast.com/s/0NerCk8vAZtNoOE3T2Iz1G?domain=people.stern.nyu.edu
Email: wgreene at stern.nyu.edu
Ph. +1.212.998.0876
Editor in Chief: Journal of Productivity Analysis
Editor in Chief: Foundations and Trends in Econometrics
Associate Editor: Economics Letters
Associate Editor: Journal of Business and Economic Statistics
Associate Editor: Journal of Choice Modeling
From wgreene at stern.nyu.edu Thu Sep 13 06:06:23 2018
From: wgreene at stern.nyu.edu (William Greene)
Date: Wed, 12 Sep 2018 16:06:23 -0400
Subject: [Limdep Nlogit List] Joint logit Model in Nlogit
In-Reply-To: <3BE1FEAD-3100-496F-B0D2-BB6F3015B109@gmail.com>
References: <3BE1FEAD-3100-496F-B0D2-BB6F3015B109@gmail.com>
Message-ID:
Minal.
It is not established in this thread what is a "joint logit model?" I do
not see the formulation
(equations) of a model that has these two models as components in something
that is
different from two separate MNL models. That has to be nailed down before
we can talk
about how to estimate it. To the point, what is correlated with what that
makes this pair of
settings different from a pair of MNL models. As it stands now, since they
are indeed just
two logit models, the way to set it up is to have the first 500 rows be the
travel mode data,
then rows 501 to 800 be be the data for the location data. Then,
SAMPLE;1-500$ followed by the travel mode analysis, then
SAMPLE; 501-800 $ followed by the residential location modeling.
If you want to follow the suggestion of folding the data into one decision,
you will have to
have a 5x3 = 15 alternative choice model. I'm not yet convinced that this
would be
plausible. Likewise, a nested logit with travel in one branch and location
in the other is also
a nonstarter, since here, every individual would be choosing both branches.
Regards
Bill Greene
On Mon, Sep 10, 2018 at 2:25 PM, minal srivastava via Limdep <
limdep at mailman.sydney.edu.au> wrote:
> Dear All,
>
> In case of forming a joint logit model for residential location choice and
> travel mode choice, how should my data format look like.
> I know for a MNL model or a Nested logit model, for each observation I
> need to convert to long data format ( depending upon the number of choice
> alternative). For e.g. for 100 observations with 5 travel mode choices, the
> total number of data input lines would be (5*100) =500.
>
> But in case of joint logit model, If I have 3 location choice alternatives
> and 5 travel mode choice alternative how should I prepare my data input
> file? Am I supposed to use two different data files, one with (3*100) 300
> input lines and other with (5*100) 500 input lines?
>
> Looking forward to your kind reply.
>
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> http://limdep.itls.usyd.edu.au
>
>
--
William Greene
Department of Economics
Stern School of Business, New York University
44 West 4 St., 7-90
New York, NY, 10012
URL: https://protect-au.mimecast.com/s/gjYFCoVzGQi46rnqc1-r89?domain=people.stern.nyu.edu
Email: wgreene at stern.nyu.edu
Ph. +1.212.998.0876
Editor in Chief: Journal of Productivity Analysis
Editor in Chief: Foundations and Trends in Econometrics
Associate Editor: Economics Letters
Associate Editor: Journal of Business and Economic Statistics
Associate Editor: Journal of Choice Modeling
From minal.crri at gmail.com Sat Sep 15 21:37:29 2018
From: minal.crri at gmail.com (minal srivastava)
Date: Sat, 15 Sep 2018 17:07:29 +0530
Subject: [Limdep Nlogit List] Joint logit model in NLOGIT5
Message-ID: <7D896DA3-3904-4360-99B8-5DA58D89DEE6@gmail.com>
Dear All,
I want to create sub models for different choice dimensions (residential location, workplace location & mode choice) and represent inter dependencies amongst these choices by using common random parameters in each utility equations. The basic idea is depicted in the utility functions for each choice dimension :
U(residential) = a1X1 + ?(residential)(Workplace)+ ? (residential)(travel mode) +?
U(workplace) = a2X2 + ?(residential)(Workplace) +?(workplace)(travel mode) +?
U(travel mode) = a3X3 + ? (residential)(travel mode) +?(workplace)(travel mode) +?
Where an and Xn are the coefficients and vector of explanatory variables. ?1, ?2 and ?3 are the error terms while, ?(residential)(Workplace), ? (residential)(travel mode) and ?(workplace)(travel mode) are the random components to capture the inter dependencies of choices made. So
My query is how can I code this in NLOGIT for the three choice processes to function as sub models in tandem with each other?
Any help would be highly appreciated.
Regards,
Minal
From minal.crri at gmail.com Sat Sep 15 21:37:29 2018
From: minal.crri at gmail.com (minal srivastava)
Date: Sat, 15 Sep 2018 17:07:29 +0530
Subject: [Limdep Nlogit List] Joint logit model in NLOGIT5
Message-ID: <7D896DA3-3904-4360-99B8-5DA58D89DEE6@gmail.com>
Dear All,
I want to create sub models for different choice dimensions (residential location, workplace location & mode choice) and represent inter dependencies amongst these choices by using common random parameters in each utility equations. The basic idea is depicted in the utility functions for each choice dimension :
U(residential) = a1X1 + ?(residential)(Workplace)+ ? (residential)(travel mode) +?
U(workplace) = a2X2 + ?(residential)(Workplace) +?(workplace)(travel mode) +?
U(travel mode) = a3X3 + ? (residential)(travel mode) +?(workplace)(travel mode) +?
Where an and Xn are the coefficients and vector of explanatory variables. ?1, ?2 and ?3 are the error terms while, ?(residential)(Workplace), ? (residential)(travel mode) and ?(workplace)(travel mode) are the random components to capture the inter dependencies of choices made. So
My query is how can I code this in NLOGIT for the three choice processes to function as sub models in tandem with each other?
Any help would be highly appreciated.
Regards,
Minal
From miq at wne.uw.edu.pl Sun Sep 16 00:21:03 2018
From: miq at wne.uw.edu.pl (=?UTF-8?Q?Miko=c5=82aj_Czajkowski?=)
Date: Sat, 15 Sep 2018 16:21:03 +0200
Subject: [Limdep Nlogit List] Joint logit model in NLOGIT5
In-Reply-To: <7D896DA3-3904-4360-99B8-5DA58D89DEE6@gmail.com>
References: <7D896DA3-3904-4360-99B8-5DA58D89DEE6@gmail.com>
Message-ID: <0b0ec5da-61da-8250-08d7-e5cb548bc525@wne.uw.edu.pl>
Dear Minal,
If I get your problem right, you could do this by stacking your data so
that for each respondent there are residential, workplace and travel
mode choices, and then the next respondent and so on. If the attributes
differ - simply use them all and replace their values with 0 if they
were not used in a particular choice (if the levels are the same for all
alternatives it does not matter anyway). And make sure you set the panel
structure right - count all choices of each respondent: residential,
workplace and travel mode. You can then easily make coefficients
choice-type specific or generic, depending on your desired specification.
One thing to probably think about when stacking your data like this and
having generic coefficients is controlling for choice-type scale
differences.
I hope this helps.
Cheers,
Mik
On 9/15/2018 1:37 PM, minal srivastava via Limdep wrote:
> Dear All,
>
> I want to create sub models for different choice dimensions (residential location, workplace location & mode choice) and represent inter dependencies amongst these choices by using common random parameters in each utility equations. The basic idea is depicted in the utility functions for each choice dimension :
>
>
> U(residential) = a1X1 + ?(residential)(Workplace)+ ? (residential)(travel mode) +?
>
> U(workplace) = a2X2 + ?(residential)(Workplace) +?(workplace)(travel mode) +?
>
> U(travel mode) = a3X3 + ? (residential)(travel mode) +?(workplace)(travel mode) +?
>
> Where an and Xn are the coefficients and vector of explanatory variables. ?1, ?2 and ?3 are the error terms while, ?(residential)(Workplace), ? (residential)(travel mode) and ?(workplace)(travel mode) are the random components to capture the inter dependencies of choices made. So
> My query is how can I code this in NLOGIT for the three choice processes to function as sub models in tandem with each other?
>
> Any help would be highly appreciated.
>
> Regards,
> Minal
>
>
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> http://limdep.itls.usyd.edu.au