search

johninpdx / RSTools

Data management tools for use with RSiena
1 stars 1 forks source link

Closing unused RODBC handles #1

Open katemills opened 7 years ago

katemills commented 7 years ago

Should on.exit(odbcClose(db)) be added to functions to prevent the many "closing unused RODBC handle" warnings that are generated when utilizing these functions?

johninpdx commented 7 years ago

Probably so. They didn’t bother me but it’s bad housekeeping, right?

One other thing. I looked at your SAOM output, and the rate parameter for period 1 (change from wave 1-3, spring to fall) is clearly a problem; it should not be more than about 8.0 (usually less). This can be (though is not invariable!) an indication that the network data has some kind of issue, or the model spec does.

I noticed also that the compchange report indicates that every actor arrived at time 1.5. This is not right, it should be time 1 for many or most of them. But ALL 89 show up as time 1.5! Fishy….. And some of them clearly should have arrived later, unless you’re using complete data—in which case you don’t need to use composition change, since there isn’t any. Usually I require that a participant has given us a survey for at least two waves in order to include them in the analysis set—but not all waves. With two, at least they contribute some change.

I need to look at the results more closely, but those are a few offhand thoughts.

More soon, and thanks. I’m not being critical, just making a few suggestions! ~j

John M. Light, Ph.D. Oregon Research Institute 1776 Millrace Drive Eugene, OR 97403 541-484-2123 (Office) 541-554-2481 (Cell)

From: Kate Mills [mailto:notifications@github.com] Sent: Monday, March 20, 2017 4:07 PM To: johninpdx/RSTools RSTools@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [johninpdx/RSTools] Closing unused RODBC handles (#1)

Should on.exit(odbcClose(db)) be added to functions to prevent the many "closing unused RODBC handle" warnings that are generated when utilizing these functions?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/johninpdx/RSTools/issues/1, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ACQQGrnIJhZWFWBN6X3mhnTb-bs5g5a8ks5rnwaCgaJpZM4MjIQQ.

katemills commented 7 years ago

Ah, sorry about that, please ignore SSD4wvALRpt.out because it was generated before I was about to update the Student Eligibility Window! I will push the latest SSD4wvALRpt.out file. Sorry about that. The compchange does actually vary across participants but for some reason I was getting a strange error before I could update the student eligibility windows and the CC was all wacky and also Wave1 was ineligible for everyone.

Kate

On Mon, Mar 20, 2017 at 4:15 PM, John Light notifications@github.com wrote:

Probably so. They didn’t bother me but it’s bad housekeeping, right?

One other thing. I looked at your SAOM output, and the rate parameter for period 1 (change from wave 1-3, spring to fall) is clearly a problem; it should not be more than about 8.0 (usually less). This can be (though is not invariable!) an indication that the network data has some kind of issue, or the model spec does.

I noticed also that the compchange report indicates that every actor arrived at time 1.5. This is not right, it should be time 1 for many or most of them. But ALL 89 show up as time 1.5! Fishy….. And some of them clearly should have arrived later, unless you’re using complete data—in which case you don’t need to use composition change, since there isn’t any. Usually I require that a participant has given us a survey for at least two waves in order to include them in the analysis set—but not all waves. With two, at least they contribute some change.

I need to look at the results more closely, but those are a few offhand thoughts.

More soon, and thanks. I’m not being critical, just making a few suggestions! ~j

John M. Light, Ph.D. Oregon Research Institute 1776 Millrace Drive Eugene, OR 97403 541-484-2123 <(541)%20484-2123> (Office) 541-554-2481 <(541)%20554-2481> (Cell)

From: Kate Mills [mailto:notifications@github.com] Sent: Monday, March 20, 2017 4:07 PM To: johninpdx/RSTools RSTools@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [johninpdx/RSTools] Closing unused RODBC handles (#1)

Should on.exit(odbcClose(db)) be added to functions to prevent the many "closing unused RODBC handle" warnings that are generated when utilizing these functions?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/john inpdx/RSTools/issues/1, or mute the threadhttps://github.com/noti fications/unsubscribe-auth/ACQQGrnIJhZWFWBN6X3mhnTb-bs5g5 a8ks5rnwaCgaJpZM4MjIQQ.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/johninpdx/RSTools/issues/1#issuecomment-287927651, or mute the thread https://github.com/notifications/unsubscribe-auth/AMvGhEWxI_mk8G5dXxF7MN1UK6O24uwEks5rnwimgaJpZM4MjIQQ .