Random Forest procedure not in Trees?

[Mike Donatello]

I have a question:

Why are the CART/CHAID/QUEST analyses available under "Classify > Tree...", but if I want a random forests analysis -- which is basically an extension of CART -- I have to plunk down another high-dollar payment for a separate SPSS module? Why is random forests not included in the "Classify" options?

I've been an SPSS user since the mid-80s, adopting it as my primary platform when SAS licenses headed for the moon. But since IBM took over, I am increasingly driven to other options (e.g., R, Python and STATA) as more cost-effective solutions. Obviously, every public business needs to perform for shareholders. I just don't understand how the current, nonsensical product strategy was hatched.

I guess this might be more of a rant than a question. If it's cluttering the list, please delete.

--

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

This might help: Depending on SPSS version (I have V23 & V24) and for both the following installed with SPSS

 

Analyze>RanFor Estimation	SPSSINC RANFOR	Estimate random forest.
Analyze>Ranfor Prediction	SPSSINC RANPRED	Compute predicted values for new data using forests from SPSSINC RANFOR.

 

 

Regards,

Will

WMB & Associates

Statistical Services

 

 

> On 8/17/2017 11:10:13 AM, Mike Donatello ([hidden email]) wrote:

> > I have a question:
>
> Why are the CART/CHAID/QUEST analyses available under "Classify > Tree...",
> but if I want a random forests analysis -- which is basically an extension
> of CART -- I have to plunk down another high-dollar payment for a separate
> SPSS module? Why is random forests not included in the "Classify" options?
>
> I've been an SPSS user since the mid-80s, adopting it as my primary
> platform when SAS licenses headed for the moon. But since IBM took over, I
> am increasingly driven to other options (e.g., R, Python and STATA) as more
> cost-effective solutions. Obviously, every public business needs to perform
> for shareholders. I just don't understand how the current, nonsensical
> product strategy was hatched.
>
> I guess this might be more of a rant than a question. If it's cluttering
> the list, please delete.
>
> --
> Mike Donatello
> [hidden email]
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

[Jon Peck]

Random forests are available as a free extension command - actually two.  SPSSINC RANFOR estimates random forests and SPSSINC RANPREP does predictions.  You might also be interested in the C5.0 trees available in the STATS C50 TREE extension command.  These all require the free R Essentials and can be installed from the Extensions menu in V24+ or the Utilities menu in V22-23.

On Thu, Aug 17, 2017 at 9:10 AM, Mike Donatello <[hidden email]> wrote:

I have a question:

Why are the CART/CHAID/QUEST analyses available under "Classify > Tree...", but if I want a random forests analysis -- which is basically an extension of CART -- I have to plunk down another high-dollar payment for a separate SPSS module? Why is random forests not included in the "Classify" options?

I've been an SPSS user since the mid-80s, adopting it as my primary platform when SAS licenses headed for the moon. But since IBM took over, I am increasingly driven to other options (e.g., R, Python and STATA) as more cost-effective solutions. Obviously, every public business needs to perform for shareholders. I just don't understand how the current, nonsensical product strategy was hatched.

I guess this might be more of a rant than a question. If it's cluttering the list, please delete.

--

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

--

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

[Mike Donatello]

Wow... I had no clue. Thanks, Jon!

On Thu, Aug 17, 2017 at 12:02 PM, Jon Peck <[hidden email]> wrote:

Random forests are available as a free extension command - actually two.  SPSSINC RANFOR estimates random forests and SPSSINC RANPREP does predictions.  You might also be interested in the C5.0 trees available in the STATS C50 TREE extension command.  These all require the free R Essentials and can be installed from the Extensions menu in V24+ or the Utilities menu in V22-23.

On Thu, Aug 17, 2017 at 9:10 AM, Mike Donatello <[hidden email]> wrote:

I have a question:

Why are the CART/CHAID/QUEST analyses available under "Classify > Tree...", but if I want a random forests analysis -- which is basically an extension of CART -- I have to plunk down another high-dollar payment for a separate SPSS module? Why is random forests not included in the "Classify" options?

I've been an SPSS user since the mid-80s, adopting it as my primary platform when SAS licenses headed for the moon. But since IBM took over, I am increasingly driven to other options (e.g., R, Python and STATA) as more cost-effective solutions. Obviously, every public business needs to perform for shareholders. I just don't understand how the current, nonsensical product strategy was hatched.

I guess this might be more of a rant than a question. If it's cluttering the list, please delete.

--

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

--

--

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

[Jon Peck]

I did not put these under the Classify menu, because they also handle regression problems.

If you look at the extension list via the menus I mentioned, you will see over 100 commands listed - all free.  I should have also mentioned STATS SVM, which is an alternative to random forests.

On Thu, Aug 17, 2017 at 12:35 PM, Mike Donatello <[hidden email]> wrote:

Wow... I had no clue. Thanks, Jon!

On Thu, Aug 17, 2017 at 12:02 PM, Jon Peck <[hidden email]> wrote:

Random forests are available as a free extension command - actually two.  SPSSINC RANFOR estimates random forests and SPSSINC RANPREP does predictions.  You might also be interested in the C5.0 trees available in the STATS C50 TREE extension command.  These all require the free R Essentials and can be installed from the Extensions menu in V24+ or the Utilities menu in V22-23.

On Thu, Aug 17, 2017 at 9:10 AM, Mike Donatello <[hidden email]> wrote:

I have a question:

Why are the CART/CHAID/QUEST analyses available under "Classify > Tree...", but if I want a random forests analysis -- which is basically an extension of CART -- I have to plunk down another high-dollar payment for a separate SPSS module? Why is random forests not included in the "Classify" options?

I've been an SPSS user since the mid-80s, adopting it as my primary platform when SAS licenses headed for the moon. But since IBM took over, I am increasingly driven to other options (e.g., R, Python and STATA) as more cost-effective solutions. Obviously, every public business needs to perform for shareholders. I just don't understand how the current, nonsensical product strategy was hatched.

I guess this might be more of a rant than a question. If it's cluttering the list, please delete.

--

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

--

--

--

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD