Browse files

New datamining ideas

  • Loading branch information...
1 parent b857f6b commit ee2e97765d0c7c0cb862d1034c820c845ffa4a9f @barttenbrinke barttenbrinke committed Dec 26, 2008
Showing with 36 additions and 0 deletions.
  1. +36 −0 TODO
@@ -15,3 +15,39 @@ Rails integration:
- World domination
+Datamining should look something like this:
+> request-log-analyzer myapp.log --interactive
+Request log analyzer builds a new database.
+Columns come from the log_parser as the LOG_LINES store all the keys that they can detect.
+Also add some extra columns like hashed_request_url etc.
+Request log analyzer then parses the logfile for its individual requests using something like the
+virtual mongrel (we need a new name for this, database_summarizer agregator? sheepdog?) combined with our
+default log parser.
+When this is done the user enters an interactive mode (like irb).
+> Filters: None
+> Total requests in database: 53232
+> $
+The user can add filters like this:
+> $ FILTER SQL ["date > ?",]
+The user will then see this:
+> Filters:
+> 1. ["date > ?",]
+> Total requests: 2120
+> $
+At any point the user can destroy filters, show the raw requests or show reports
+The request remaining after the filter chain will then be processed through the summarizer and then trough the
+output templates, generating reports specificly for the selected dataset.
+Partials should also be possible

0 comments on commit ee2e977

Please sign in to comment.