About mathias

mathias · ‎10-03-2014

This was indeed what I intended. But I couldn't ask for the whole solution without searching more by myself . And your solution is shorter than what I have so far, Thanks data scannedStrings; infile data3 lrecl=32760 scanover; do cases=1 to 1001 /*until(value='')*/; input @'new Date(' dateString $11. @; input @'},{"v":' valueString $4. @; output; *put '---' cases ' ---' valueString; end; drop cases; stop; run; data GoogleTrends; format date IS8601DA10. ; set scannedStrings; year=scan(dateString,1,','); month=scan(dateString,2,',')+1; day=scan(scan(dateString,3,','),1,')'); date=mdy(month,day,year); value=input(scan(valueString,1,','),best5.); search="&search"; diff=date-lag(date); keep date value search; run; I still havn't found a solution to google's request limit. I could not run your code twice.

mathias · ‎10-01-2014

Is it possible to test if the filename was incorrectly fetched ? like if the text contains <!DOCTYPE html>, just don't use it filename mydata url 'http://www.google.com/trends/fetchComponent?q=tekenbeet&cid=TIMESERIES_GRAPH_0&export=3&geo=BE&date=today%2024-m'; /* if contains <!DOCTYPE html> then do not copy mydata in gTrends.json */ data _null_; infile mydata ; file '...\gTrends.json'; input; put _infile_; run; filename copydata '...\gTrends.json';

mathias · ‎10-01-2014

I played with the filename and it seems to work now, I need to use a local copy of the text. Thank you very much !

mathias · ‎10-01-2014

Thanks for your help Is this code working for you ? On my side, it just gives an empty dataset with 3 columns (f cases v). filename mydata url 'http://www.google.com/trends/fetchComponent?q=sas&cid=TIMESERIES_GRAPH_0&export=3&geo=BE&date=today%2024-m'; data XXXX; infile mydata; Length f $3; Do cases =1 to 1000 until( f= ' ' ); input @'"v":' v $8. @'"f":"' f $3. @; output; End; Stop; run;

mathias · ‎09-30-2014

Hello, I need to import google trends data from an url http://www.google.com/trends/fetchComponent?q=sas&cid=TIMESERIES_GRAPH_0&export=3&geo=BE&date=today%2024-m I found on the web a method using a data step : filename mydata url 'http://www.google.com/trends/fetchComponent?q=sas&cid=TIMESERIES_GRAPH_0&export=3&geo=BE&date=today%2024-m' ; data XXXX; infile mydata scanover; input @'"v":' v $8. @'"f":"' f $3. ; run; However, because the file has all the records on one line, the data step create only one record. If I manually enter carriage returns in the json file, I get the intended result. Does anyone has a solution for this particular problem ? Or maybe another method to parse the date from this json text file ? Many thanks,

mathias · ‎09-12-2014

Thanks a lot, I tried what you suggested. I got my stp to output {"success":1} (removed % in your data step + added %stepbegin right after it) However, what happens is always : //if fails In the console I have this error : XMLHttpRequest cannot load <formURL>. No 'Access-Control-Allow-Origin' header is present on the requested resource. Origin '<MyWebsiteURL>' is therefore not allowed access. And here is what I found on the subject : http://stackoverflow.com/questions/19821753/jquery-xml-error-no-access-control-allow-origin-header-is-present-on-the-req#19821851 But I have no idea how to enable CORS on the STP server.

mathias · ‎09-12-2014

I'm using jQuery, yes. can you tell me little bit more ? Thanks

mathias · ‎09-03-2014

Maybe the stored process can be run in an invisible iframe in my website, this would make the process completely invisible to users. Browsers may not like invisible iframes though. I'll try this.

mathias · ‎09-03-2014

Hi, I would like to register email subscriptions from a static website (only html and javascript) I have a simple webform with 3 values that need to be registerd. I was planning to do it with a SAS STP and store the 3 values in a dataset. It's working but my problem is that UX is really bad : - when a user click on submit, the browser goes to the link of the STP, closing the current website (can be solved with newwindow) - the stp does it's job, it registers the subscription into a dataset. - user sees a gray screen and is left there. How is it possible to make this more friendly ? Ideally the user would click on submit, would be redirected to the same website and would be shown a confirmation dialog (passed with an URL parameter like "?subscriptionIsConfirmed=1" ) Is there a way to run the STP in the background ? to prevent it of opening a window ? to close the STP window ? to redirect to a specific URL ? Thank you for your help !

mathias · ‎08-14-2014

Thank you, Yes I'm using EG. And checking "Customize results formats (...)" did the job. Is there any way of doing this with code ? Thanks !

mathias · ‎08-14-2014

I'm not sure to understand everything but if you have a character variable with an observation containing '505546433549' then in your querry filter you should be able to write VariableX Equal to 505546433549 (without quotes) Enclose values in quotes (check this) OR "505546433549" (with quotes, double or single) Enclose values in quotes (unchecked) Does this help ?

mathias · ‎08-14-2014

Hello, I have trouble inserting the title of an sgplot into the png image, even with ods gtitle option. (The SAS/GRAPH gmap has titles inserted into the png, but the ODS GRAPHICS sgplot does not.) Any idea what I might be missing ? Thank you ! Here the relevant sample code. ods html path="&graphPath" device=png image_dpi=100 file='file.html' body='body.html' style=&style gtitle gfootnote ; (...) /* ODS GRAPHICS */ ods graphics on / reset noborder imagename="bigHist_&Subject" height=200mm width=323.6mm ; title "%trim(%sysfunc(putc(&Subject,code2displayName.)))"; title1 "%trim(%sysfunc(putc(&Subject,code2displayName.)))"; proc sgplot data=bigHsitogramDS ; vbar Week / response=CasesYear0 barwidth=1 nooutline transparency=.2 name="vbar" legendlabel="Present" ; vline Week / response=CasesYear12 stat=mean lineattrs=(pattern=solid thickness=2) transparency=.2 name="vline2" legendlabel="Past (mean of 2 years smoothed)" ; refline %sysfunc(week(&lastInterval,W)) / axis=x label="This Week"; keylegend / ; yaxis label="Weekly Reported Cases" integer min=0 grid /*values=(0 to 10 by 10)*/; xaxis display=(nolabel) type=discrete ; run; (...) /* SAS/GRAPH */ goptions reset=all goutmode=replace device=png cback=white noborder hsize=12cm vsize=4cm; * define the choro map color ; pattern color=CXD6E1F1; * create the graph ; title; proc gmap map=common.BEMAP_PROVINCES data=common.BEMAP_PROVINCES anno=annotateDS density=1 ; id NIS1_5; choro DENSITY / levels=1 statistic=first missing coutline=black woutline=1 nolegend name="miniMap_&Subject"; run;quit; * reset ; pattern; (...) ods html close;

mathias · ‎06-18-2014

This the best I can do with the HIGHLOW + data ticks are much better - barwidth=1 does not work, I can play with intervalbarwidth but its an absolute value, thus not dynamic - group colors are inverted, whatever comes first determines the color Code proc sgplot data=HIGHLOWPLOT ; *where DateUsedForStatistics>'01jan2013'd; highlow x=DateUsedForStatistics low=low high=high / group=isAugmenting name='a' type=bar barwidth=1 discreteoffset=0 fill nooutline transparency=.2 /*intervalbarwidth=3*/ ; *keylegend 'a' / title='Virus' location=inside position=topright across=1; xaxis type=time interval=semiyear offsetmin=0 offsetmax=0; yaxis grid display=(NOTICKS); run; This is the best I can do with the NEEDLE same advantages and disadvantages than highlow - lines are drawn in the legend Code proc sgplot data=HIGHLOWPLOT ; *where DateUsedForStatistics>'01jan2013'd; needle x=DateUsedForStatistics y=Cases / group=isAugmenting lineattrs=(thickness=2 pattern=solid) transparency=.2 ; *keylegend 'a' / title='Virus' location=inside position=topright across=1; xaxis type=time interval=semiyear offsetmin=0 offsetmax=0; yaxis grid display=(NOTICKS); run;

mathias · ‎06-18-2014

I don't know what is the cause, but when I ran the program again today with DiscreteOpts=(TickValueFitPolicy=thin) , it was correct : thank you ! Is there a way to force the intervals to be at the beginning of each year ?

mathias · ‎06-18-2014

Thank you Sanjay TickValueFitPolicy=thin seems to work but it modifies the chart : (other TickValueFitPolicy values I picked from examples i found, produce errors) I'll have a look at the article and try to reproduce that.

Re: use infile scanover when only one line in JSON file

Re: use infile scanover when only one line in JSON file

Re: use infile scanover when only one line in JSON file

Re: use infile scanover when only one line in JSON file

use infile scanover when only one line in JSON file

Re: background invisible stored process

Re: background invisible stored process

Re: background invisible stored process

background invisible stored process

Re: embed titles in png with ods graphics

Re: Proper query syntax for a "character-type" variable"

embed titles in png with ods graphics

Re: GTL barchart with too much labels for time xaxis

Re: GTL barchart with too much labels for time xaxis

Re: GTL barchart with too much labels for time xaxis