group by 分组（集团通过分组）.docVIP

group by 分组（集团通过分组） Fifth section group by grouping application First of all, how do groups work?. For instance： Table: temp1 has the following records: BH sl AAA 1 AAA 4 CCC 2 BBB 5 AAA 9 BBB 7 Now count how many temp1 numbers there are, and what is the total number of all the numbers. Obviously, this is the use of grouping. Sele, BH, sum (SL), as, total_sl; From temp1; Grou by bh; Orde by bh; Into curs temp2 The operation of the order is like the counting of votes in a ballot. It is exactly what kind of technology to store intermediate results I dont know, but this is not important, I assume it using temporary tables now (actually I think unlikely), do not affect the results, but also easy to understand. 1, scan the records one by one. Every time a new number, on the establishment of a temporary table for this number (again, using temporary tables just assume that layout, my method really I dont know), then put the temporary table records into the corresponding to. If its not a new number, put the record directly into the corresponding temporary table. 2, all the records after the scan, the corresponding number of temporary tables also produced, and now is for each temporary table using sum () for statistics. For example, the temporary table named AAA is cursor1, and the content of that is..: Aaa 1 Aaa 4 Aaa 9 Now sum () statistics, the effect is equivalent to sele, BH, sum (SL), as, total_sl, from, cursor1. Each numbered temporary table performs such sum () statistics, and then combines the statistical results of each number together, just like using Union, and the final result is..: AAA 14 CCC 2 BBB 12 3, and then sort the BH, and the result is: AAA 14 BBB 12 CCC 2 Looked at the above example, and then look at the following command, presumably also understandable: Sele, cpk.cpbh, cpk.cpmc, sum (IIF (isnull (thdmx.thsl), 0, thdmx.thsl) as, total_thsl; From, Cpk, left, join, thdmx; On cpk.cpbh=thdmx.cpbh; Grou by cpk.cpbh; Orde by cpk.cpbh; Into curs temp1 The pro