If I create a view and then a mining structure run it everything works fine BUT if I then alter the view (e.g change a filter setting in a named query) or the actual contents of the database, the OLD contents persist.
Using the view refresh says there are no changes (which is correct as I don't change any column names etc.). If I remove Keep Training Cases, this fixes it, but for Time Series, it then gives an error, so you have to go back and re-select it... is there a direct option to clear caches?
The dependency settings aren't sensitive to non-destructive changes in the DSV. You can simply do a ProcessFull on the mining structure to reprocess from the data.
There is an option ProcessClearStructureOnly (something like that), that clears the training cases from the cache.
TimeSeries requires the cache or the viewer will not work - you can likely drop the cache using the process method above, but you will get errors when trying to view the time series models (at least the chart part)
No comments:
Post a Comment