From: Paul on 19 Jan 2010 23:32 I have a dataset with the following columns: This is a one to many relationship between the health authority to city and one to many relationship between the city and hospital column_group: health authority, city and hospital (three possible groupings for each of the row in the dataset) column_health_authority_name: it displays the name of the health authority to the row with the group "health authority" and it displays null to the "city" and "hospital" group column_city_name: it displays the city name to the row with the group "city" and "hospital" and it displays null to the row with the group "health authority" column_hospital_name: it displays the name of the hospital and displays null to the row with the group "health authority" and "city" column_maesure_1: some kind of measure applicable to the three groupings. column_measure_2: some kind of measure applicable to the three groupings. and so on for half a dozen measure columns I want to create a table report with the "health authority" as the highest grouping and "city" as the second groupng and "hospital" as the lowest grouping. Initially the report shows only the "health authority" data and user can drill down to the next grouping level (or drill up back to the next grouping) and so on drill down to the "hosppital" grouping. I have tried all different grouping combinations to achieve this with no luck. I created three groupings in the report but not sure how to group this to make it work. I just want to show the data the way it is in the dataset and no further aggregation in the report is required. Thanks.
From: Uri Dimant on 20 Jan 2010 02:50 Paul Can you please post sample data + table structure + an expected result for testing? "Paul" <paul_mak(a)shaw.ca> wrote in message news:eF5wTmYmKHA.5728(a)TK2MSFTNGP06.phx.gbl... >I have a dataset with the following columns: > > This is a one to many relationship between the health authority to city > and one to many relationship between the city and hospital > > column_group: health authority, city and hospital (three possible > groupings for each of the row in the dataset) > column_health_authority_name: it displays the name of the health authority > to the row with the group "health authority" and it displays null to the > "city" and "hospital" group > column_city_name: it displays the city name to the row with the group > "city" and "hospital" and it displays null to the row with the group > "health authority" > column_hospital_name: it displays the name of the hospital and displays > null to the row with the group "health authority" and "city" > column_maesure_1: some kind of measure applicable to the three groupings. > column_measure_2: some kind of measure applicable to the three groupings. > > and so on for half a dozen measure columns > > I want to create a table report with the "health authority" as the highest > grouping and "city" as the second groupng and "hospital" as the lowest > grouping. Initially the report shows only the "health authority" data and > user can drill down to the next grouping level (or drill up back to the > next grouping) and so on drill down to the "hosppital" grouping. I have > tried all different grouping combinations to achieve this with no luck. I > created three groupings in the report but not sure how to group this to > make it work. I just want to show the data the way it is in the dataset > and no further aggregation in the report is required. Thanks. > >
|
Pages: 1 Prev: How to access dataset from embedded into RDL code (urgent) Next: SQL 2005 PDF Export |