I am attempting to run queries on a table and then do some summary calculations on the results. However, I seem to be stuck with a results format which can't be referenced easily by columns nor have calculations run on it. The code below may be unnecessarily complicated but I was unable to get slices to work properly on non-contiguous time periods with table.read(start,stop,step).
<snippet of code> tser2 = [x.nrow for x in table1.where(foo_condition)] # tser3 = map(int,tser2) tser4 = numpy.array(tser3) tser5 = tser4 + 1 tser6 = tser4 + 2 month1 = table1.readCoordinates(tser4) month2 = table1.readCoordinates(tser5) </snippet ends> Essentially I am taking a CSV file, converting the lines to numpy arrays and loading them into tables. Upon running a query, I retrieve the coordinates so I can then increment the rows forward by 1 and 2 periods and collect the observations. However I am left with an ndarray with each line type as numpy void. Running calculations on this type, (i.e. cumsum by row) causes an error "cannot perform accumulate with flexible type". Happened to find an earlier discussion by Mr. Stephen Simmons whereby he converted to numpy.array post the table.read(). So I am unsure as to the right path to pursue. I was under the impression, when a numpy array was imported into the table, one should get a numpy array as a result query. The mis-guided read.coordinates() exercise must be screwing this up Unfortunately I can't include all the code due to the proprietary nature of the research I am doing. Would appreciate any tips and/or tricks however. Thank you for your valuable time. Best regards, -- Vince Fulco ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Pytables-users mailing list Pytables-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/pytables-users
