Hi Ashok,
It would help if post an image from your table view with your dimensional tables and your three separate fact tables. There will be more than one way of doing this. My approach is usually to concatenate your fact tables but this may involve moving some of your fields around.
Suppose you have two fact tables with a one to many relationship such as order header and order items with a number of dimensions associated with your header table, such as Customer and others, the two fact tables associate on a field like OrderID.
"Move" the Customer field into the Items table by something like
MappingCustomerRef:
Mapping
LOAD
OrderID,
CustID
Resident OrderHeaders;
Left Join(Items)
LOAD DISTINCT
OrderID,
ApplyMap('MappingCustomerRef',OrderID,'Missing') as CustID
Resident Items;
DROP Field CustID from OrderHeaders;
Now the Customer dimensional table will associate with the records in the Items table.
Do this for all dimensions associated with the header table until the Header table has only one association, that is with the Items table on the field OrderID. At this point you can concatenate the two tables.
A final point: star schemas are all very well and easy to understand but I wouldn't get obsessed with always obtaining that model. Sometimes there is a high price to pay in memory usage when creating large fact tables and performance may in fact suffer.
Cheers
Andrew
