Technology Tales

ERROR: Ambiguous reference, column xx is in more than one table.

Sometimes, SAS messages are not all that they seem, and a number of them are issued from PROC SQL when something goes awry with your code. In fact, I got a message like the above when ordering the results of the join using a variable that didn't exist in either of the datasets that were joined. This type of thing has been around for a while (I have been using SAS since version 6.11, and it was there then) and it amazes me that we haven't seen a better message in more recent versions of SAS; it was SAS 9.2 where I saw it most recently.

proc sql noprint;
    select a.yy, a.yyy, b.zz
        from a left join b
            on a.yy=b.yy
                order by xx;
quit;

Changing from to Nvidia Graphics Drivers on Linux Mint Debian Edition 64-bit

One way of doing this is to go to the Nvidia website and download the latest file from the relevant page on there. Then, the next stage is to restart your PC and choose rescue mode instead of the more usual graphical option. This drops you onto a command shell that is requesting your root password. Once this is done, you can move onto the next stage of the exercise. Migrate to the directory where the *.run file is located and issue a command similar to the following:

bash NVIDIA-Linux-x86_64-295.40.run

The above was the latest file available at the time of writing, which may have changed by the time that you read this. If the executable asks to modify your X configuration file, I believe that the best course is to let it do that. Editing it yourself or running nvidia-xconfig are alternative approaches if you so prefer.

Since proprietary Nvidia drivers are included in the repositories for Linux Mint Debian Edition, that may be a better course of action since you will get updates through normal system update channels. Then, the course of action is to start by issuing the following installation commands:

sudo apt-get install module-assistant
sudo apt-get install nvidia-kernel-common
sudo apt-get install nvidia-glx
sudo apt-get install kernel-source-NVIDIA
sudo apt-get install nvidia-xconfig

Once those have completed, issuing the following in turn will complete the job ahead of a reboot:

sudo m-a a-i nvidia
sudo modprobe nvidia
sudo nvidia-xconfig

If you reboot before running the above like I did, you will get a black screen with a flashing cursor instead of a full desktop because X failed to load. Then, the remedy is to reboot the machine and choose the rescue mode option, provide the root password and issue the three commands (at this point, the sudo prefix can be dropped because it's unneeded) then. Another reboot will see order restored and the new driver in place. Running the following at that point will do a check on things, as will be the general appearance of everything:

glxinfo | grep render

Synchronising package selections between Linux Mint and Linux Mint Debian Edition

To generate the package list on the GNOME version of Linux Mint, I used the Backup Tool. It simply was a matter of using the Backup Software Selection button and telling it where to put the file that it generates. Alternatively, dpkg can be used from the command line like this:

sudo dpkg --get-selections > /backup/installed-software.txt

After transferring the file to the machine with Linux Mint Debian Edition, I tried using the Backup Tool on there too. However, using the Restore Software Selection button and loading the required only produced an irrecoverable error. Therefore, I set to looking around the web and found a command line approach that did the job for me.

The first step is to load the software selection using dpkg by issuing this command (it didn't matter that the file wasn't made using the dpkg command, though I suspect that's what the Linux Mint Backup Tool was doing that behind the scenes):

sudo dpkg --set-selections < /backup/installed-software.txt

Then, I started dselect and chose the installation option from the menu that appeared. The first time around, it fell over but trying again was enough to complete the job. Packages available to the vanilla variant of Linux Mint but not found in the LMDE repositories were overlooked as I had hoped, and installation of the extra packages had no impact on system stability either.

sudo dselect

Apparently, there is an alternative to using dselect that is based on the much used apt-get command, but I didn't make use of it so cannot say more:

sudo apt-get dselect-upgrade

All that I can say is that the dpkg/dselect combination did what I wanted, so I'll keep them in mind if ever need to synchronise software selections between two Debian-based distributions in the future again. While the standard edition of Linux Mint may be based on Ubuntu rather than Debian, Ubuntu is itself based on Debian. Thus, the description holds here.

Uninstalling VirtualBox Guest Additions on a Linux Guest OS

Within the last few days, I updated my Linux Mint Debian Edition virtual machine installation to Update 4. Between not following the instructions so closely and problems with the update server, a re-installation preceded the update itself. When all was done, no desktop environment appeared, and issuing the startx command revealed that it was one of the VirtualBox drivers that was the cause of the problem. With my being unable to see any files on the VirtualBox virtual CD, something else needed doing and the executing following command (replacing [VboxAddonsFolder] with VBoxGuestAdditions-4.1.12 in my case, but it is different for each VirtualBox version) resolved the situation:

/opt/[VboxAddonsFolder]/uninstall.sh

When it was complete, a scrambled desktop began to appear, so a reboot was to set things to rights. Then, I could set to looking at what Update 4 had brought to Linux Mint Debian Edition.

Calculating geometric means using SAS

Recently, I needed to calculate geometric means after a break of a number of years and ended racking a weary brain until it was brought back to me. In order that I am not in the same situation again, I am recording it here and sharing it is always good too.

The first step is to take the natural log (to base e or the approximated irrational value of the mathematical constant, 2.718281828) of the actual values in the data. Since you cannot have a log of zero, the solution is either to exclude those values or substitute a small value that will not affect the overall result, as is done in the data step below. In SAS, the log function uses the number e as its base, so you need to use the log10 equivalent when base 10 logs are needed.

data temp;
    set temp;
    if result=0 then _result=0.000001;
    else _result=result;
    ln_result=log(_result);
run;

Next, the mean of the log values is determined, and you can use any method of doing that so long as it gives the correct values. PROC MEANS is used here but PROC SUMMARY (identical to MEANS except it defaults to data set creation while that generates output by default; for that reason, we need to use the NOPRINT option here), PROC UNIVARIATE or even the MEAN function in PROC SQL.

proc means data=temp noprint;
    output out=mean mean=mean;
    var ln_result;
run;

With the mean of the log values obtained, we have to take the exponential of the obtained value(s) using the EXP function. This returns values of the same magnitude as in the original data, using the formula emean.

data gmean;
    set mean;
    gmean=exp(mean);
run;

Creating placeholder graphics in SAS using PROC GSLIDE for when no data are available

Recently, I found myself with a plot to produce, but there were no data to be presented, so a placeholder output was needed. For a listing or a table, this is a matter of detecting if there are observations to be listed or summarised and then issuing a placeholder listing using PROC REPORT if there are no data available. Using SAS/GRAPH, something similar can be achieved using one of its curiosities.

In the case of SAS/GRAPH, PROC GSLIDE looks like the tool to user for the same purpose. The procedure does get covered as part of a SAS Institute SAS/GRAPH training course, but they tend to gloss over it. After all, there is little reason to go creating presentations in SAS when PowerPoint and its kind offer far more functionality. However, it would make an interesting tale to tell how GSLIDE became part of SAS/GRAPH in the first place. Its existence makes me wonder if it pre-exists the main slideshow production tools that we use today.

The code that uses PROC GSLIDE to create a placeholder graphic is as follows (detection of the number of observations in a SAS dataset is another entry on here):

proc gslide;
    note height=10;
    note j=center "No data are available";
run;
quit;

PROC GSLIDE is one of those run group procedures in SAS so a QUIT statement is needed to close it. The NOTE statements specify the text to be added to the graphic. The first of these creates a blank line of the required height for placing the main text in the middle of the graphic. It is the second one that adds the centred text that tells users of the generated output what has happened.

Creating waterfall plots in SAS using PROC GCHART

Recently, I needed to create a waterfall plot couldn't use PROC SGPLOT since it was incompatible with publishing macros that use PROC GREPLAY on the platform that I was using; SGPLOT doesn't generate plots in SAS catalogues but directly creates graphics files instead. Therefore, I decided that PROC GCHART needed to be given a go, and it delivered what was needed.

The first step is to get the data into the required sort order:

proc sort data=temp;
    by descending result;
run;

Then, it is time to add an ID variable for use in the plot's X-axis (or midpoint axis in PROC GCHART) using an implied value retention to ensure that every record in the dataset had a unique identifier:

data temp;
    set temp;
    id+1;
run;

After that, axes have to be set up as needed. For instance, the X-axis (the axis2 statement below) needs to be just a line with no labels or tick marks on there and the Y-axis was fully set up with these, turning the label from vertical to horizontal as needed with the ANGLE option controlling the overall angle of the word(s) and the ROTATE option dealing with the letters, and a range declaration using the ORDER option.

axis1 label=none major=none minor=none value=none;
axis2 label=(rotate=0 angle=90 "Result") order=(-50 to 80 by 10);

With the axis statements declared, the GCHART procedure can be defined. Of this, the VBAR statement is the engine of the plot creation, with the ID variable used for the midpoint axis and the result variable used as the summary variable for the Y-axis. The DISCRETE keyword is needed to produce a bar for every value of the ID variable, or GCHART will bundle them by default. Next, references for the above axis statements (MAXIS option for midpoint axis and AXIS option for Y-axis) are added, and the plot definition is complete. One thing that has to be remembered is that GCHART uses run group processing, so a QUIT statement is needed at the end to close it at execution time. This feature has its uses and appears in other procedures too, though SAS procedures generally are concluded by a RUN statement.

proc gchart data=temp;
    vbar id / sumvar=result discrete axis=axis2 maxis=axis1;
run;
quit;

Do we need dongles anymore?

Recent exposure to both the HTC Wildfire S and the HTC Desire S has me wondering if we need mobile broadband dongles any more. The reason for my asking this is that both Android devices can act as mobile Wi-Fi hubs, and they work very well as these too. Even the dedicated T-Mobile mobile Wi-Fi hub that I picked up in the closing months of 2011 now looks a little obsolete, though it retains a cost advantage in its favour.

In the case of both HTC phones, it thankfully is possible to use high security encryption and a pass key too. However, it is best to change the default key before any activation of Wi-Fi signal, if only to ensure that you don't end up with a very nasty bill. The Wi-Fi Hotspot App has all the settings that you need and up to five connections can be supported at a time, just like that T-Mobile hub that I mentioned earlier.

That replaced a defunct dongle from the same company; the USB connection appeared to have failed and was ailing for a while. Now, it might be that the more transparent use of an actual mobile phone might be usurping that as well, especially when I have been wondering why it has been doing so well with internet connectivity while using it on the move. This voyage into the world of what smartphones and similar devices can do is throwing up its share of surprises as I go along.

Smoother use of more than one SAS DMS session at a time

Unless you have access to SAS Enterprise Guide, being able to work on one project at a time can be a little inconvenient. It is possible to open up more than one Display Manager System (DMS, the traditional SAS programming interface) session at a time only to get a pop-up window for SAS documentation for the second and subsequent sessions. You don't get your settings shared across them, either, while also losing any changes to session options after shutdown.

The cause of both of the above is the locking of the SASUSER directory files by the first SAS session. However, it is possible to set up a number of directories and set the -sasuser option to point at different ones for different sessions.

On Windows, the command in the SAS shortcut becomes:

C:\Program Files\SAS\SAS 9.1\sas.exe -sasuser "c:\sasuser\session 1\"

On UNIX or Linux, it would look similar to this:

sas -sasuser "~/sasuser/session1/"

Since the "session1" in the folder paths above can be replaced with whatever you need, you can have as many as you want too. It might not seem much of a need but synchronising the SASUSER folders every now and again can give you a more consistent set of settings across each session, all without intrusive pop up boxes or extra messages in the log too.

Changing to web fonts

While you can add Windows fonts to Linux installations, I have found that their display can be flaky to say the least. Linux Mint and Ubuntu display them as sharp as I'd like, but I have struggled to get the same sort of results from Arch Linux, while I am not so sure about Fedora or openSUSE either.

This led me to explore web fonts for my websites, with Google Web Fonts meeting my needs through options like Open Sans and Arimo. There have been others with which I have dallied, such as Droid Sans, but these are the ones on which I have settled for now. Both are in use on this website now, and I added calls for them to the web page headers using the following code (lines are wrapping due to space constraints):

<link href="http://fonts.googleapis.com/css?family=Open+Sans:300italic,400italic,600italic,700italic,400,300,600,700" rel="stylesheet" type="text/css">
<link href='http://fonts.googleapis.com/css?family=Arimo:400,400italic,700,700italic' rel='stylesheet' type='text/css'>

With those lines in place, it then is a matter of updating font-family and font declarations in CSS style sheets with "Open Sans" or "Arimo" as needed, while keeping alternatives defined in case the Google font service goes down for whatever reason. A look at a development release of the WordPress Twenty Twelve theme caused me to come across Open Sans and I like it for its clean lines and Arimo, which was found by looking through the growing Google Web Fonts catalogue, is not far behind. Looking through that catalogue now causes for me a round of indecision since there is so much choice. For that reason, I think it's better to be open to the recommendations of others.