Help to deicde analysis

[samuelsr]

Hello people, Hope everyone is safe and sound with COVID 19 rampage.

I have a birth cohort data, and I want to establish a novel link between risk factor and an outcome 

say the risk factors are Breast milk, alternate feed and combination towards obesity. don't decipher into the biological plausibility of the variables this is an eg. 

but one of the factors (biologically sig) is not significant because of high collinearity. But if I want to find the true link and avoid this issue how do I go about? 

desperate for help as I don't have time guys. 

regards

===================== 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

Maybe explore Shapley Value regression.

Am Samstag, 4. April 2020, 18:39:10 MESZ hat Mario Giesel <[hidden email]> Folgendes geschrieben:

Maybe explore Shapley Value regression.

Am Samstag, 4. April 2020, 18:12:34 MESZ hat samuel raj srinivasan <[hidden email]> Folgendes geschrieben:

Hello people, Hope everyone is safe and sound with COVID 19 rampage.

I have a birth cohort data, and I want to establish a novel link between risk factor and an outcome 

say the risk factors are Breast milk, alternate feed and combination towards obesity. don't decipher into the biological plausibility of the variables this is an eg. 

but one of the factors (biologically sig) is not significant because of high collinearity. But if I want to find the true link and avoid this issue how do I go about? 

desperate for help as I don't have time guys. 

regards

===================== 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

[samuelsr]

can you let me know how to do that pls. I am a self taught biostatistician.

On Sat, Apr 4, 2020 at 10:10 PM Mario Giesel <[hidden email]> wrote:

Maybe explore Shapley Value regression.

Mario Giesel

Munich, Germany

Am Samstag, 4. April 2020, 18:39:10 MESZ hat Mario Giesel <[hidden email]> Folgendes geschrieben:

Maybe explore Shapley Value regression.

Mario Giesel

Munich, Germany

Am Samstag, 4. April 2020, 18:12:34 MESZ hat samuel raj srinivasan <[hidden email]> Folgendes geschrieben:

Hello people, Hope everyone is safe and sound with COVID 19 rampage.

I have a birth cohort data, and I want to establish a novel link between risk factor and an outcome 

say the risk factors are Breast milk, alternate feed and combination towards obesity. don't decipher into the biological plausibility of the variables this is an eg. 

but one of the factors (biologically sig) is not significant because of high collinearity. But if I want to find the true link and avoid this issue how do I go about? 

desperate for help as I don't have time guys. 

regards

Dr. S.Samuel Raj,

Associate Professor,

Department of Public Health Dentistry

Saveetha Dental College

Chennai, Tamil Nadu, India 

Ph- 91- 7010395428

https://orcid.org/0000-0002-7344-4476

https://www.researchgate.net/profile/Srinivasan_Samuel

===================== 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

===================== 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]

The Shapley value method is one of many methods for variable selection in regression, but it's pretty good IMO.  In SPSS Statistics, it is available in the STATS RELIMP (relative importance) extension command, which also provides five other importance measures.  You can install it from the Extensions > Extension Hub menu, but it also requires R and the R Essentials.  You can get details about those here:

https://developer.ibm.com/predictiveanalytics/downloads/ 

Note: the migration to a new analytics community site has made this link rather harder to find at the moment, so save this link if you might need it in the future. 

On Sat, Apr 4, 2020 at 10:47 AM samuel raj srinivasan <[hidden email]> wrote:

can you let me know how to do that pls. I am a self taught biostatistician.

Dr. S.Samuel Raj,

Associate Professor,

Department of Public Health Dentistry

Saveetha Dental College

Chennai, Tamil Nadu, India 

Ph- 91- 7010395428

https://orcid.org/0000-0002-7344-4476

https://www.researchgate.net/profile/Srinivasan_Samuel

On Sat, Apr 4, 2020 at 10:10 PM Mario Giesel <[hidden email]> wrote:

Maybe explore Shapley Value regression.

Mario Giesel

Munich, Germany

Am Samstag, 4. April 2020, 18:39:10 MESZ hat Mario Giesel <[hidden email]> Folgendes geschrieben:

Maybe explore Shapley Value regression.

Mario Giesel

Munich, Germany

Am Samstag, 4. April 2020, 18:12:34 MESZ hat samuel raj srinivasan <[hidden email]> Folgendes geschrieben:

Hello people, Hope everyone is safe and sound with COVID 19 rampage.

I have a birth cohort data, and I want to establish a novel link between risk factor and an outcome 

say the risk factors are Breast milk, alternate feed and combination towards obesity. don't decipher into the biological plausibility of the variables this is an eg. 

but one of the factors (biologically sig) is not significant because of high collinearity. But if I want to find the true link and avoid this issue how do I go about? 

desperate for help as I don't have time guys. 

regards

Dr. S.Samuel Raj,

Associate Professor,

Department of Public Health Dentistry

Saveetha Dental College

Chennai, Tamil Nadu, India 

Ph- 91- 7010395428

https://orcid.org/0000-0002-7344-4476

https://www.researchgate.net/profile/Srinivasan_Samuel

===================== 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

===================== 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