I'm creating a statistics system for events. When an event has passed,
an admin can enter statistics for that passed event. Every user
attending the event and the event itself have statistics. The problem
is that depending on the type of the event, the type of statistics for
both users and event can change. eg: for one event the number of
remarks and time present will be registered for another event the name
of the company he represents
I'm struggling on how to implement this especially database wise. I
can ofcourse create a table that contains all possible statistics data
and than only fill in the rows that matter for that event. the
downside is that I lose a lot of storage since a lot of rows are left
blanc.
is that depending on the type of the event, the type of statistics for
both users and event can change. eg: for one event the number of
remarks and time present will be registered, for another event the name
of the company he represents
You could approach this with a key-key value type of table definition.
STATISTIC:
EVENT_ID
STAT
STAT_TYPE
STAT_STRING
EVENT_ID can link back to the event
STAT is the name of each statistic, like REMARK_COUNT, TIME_PRESENT, or
COMPANY_NAME, however many you'd like to keep (I'd probably keep the
valid list in another table).
STAT_TYPE is the data type of the value "stored", which directs you to
which of the remaining fields holds the valid value for this record.
STAT_STRING is the stringified version of the data.
You could record any number of different statistics for any event, it's
row-heavy, column-light, and has some of its own issues, but it's
workable.
I'm sure there's a better approach, but it's been like 10 years since I
worked on an issue where I wanted to store an arbitrary collection of
attributes.
