Pattern Discovery an Example with Sequential Patterns Agrawal

Pattern Discovery: What if contextual information is available? Rosa Frédérick François Bruno Pascal Osmar

Slides: 8

Download presentation

Pattern Discovery: an Example with Sequential Patterns [Agrawal and Srikant 1996] Rosa Frédérick François Bruno Pascal Osmar Sandra Maguelonne Julien Alberto <( )( )> <( )( )( )> <( <( )( )( )( <( )( )> )> “Rosa had lunch, then later put on some sunglasses while going to the beach” <( Support Frequency )( )> “How many people? ” 6 “What proportion? ” 6/10 A frequent behavior is followed by at least 50% of people <( )( )> is a frequent sequence!

Pattern Discovery: What if contextual information is available? Rosa Frédérick François Bruno Pascal Osmar Sandra Maguelonne Julien Alberto Contextual Information <( )( )> <( )( )( )> <( <( )( )( )( <( )( )> )> Working or Not working

Pattern Discovery: What if contextual information is available? Rosa Frédérick François Bruno Pascal Osmar Sandra Maguelonne Julien Alberto Contextual Information <( )( )> )( )> <( )( )( )( <( )( Sunny <( <( <( Working )( or Not working or Rainy )> is frequent but: • 6 out of 6 working )> )( )> )> • none not working Problem: larger proportion of working people

Pattern Discovery: What if contextual information is available? Rosa Frédérick François Bruno Pascal Osmar Sandra Maguelonne Julien Alberto Contextual Information <( )( )> )( )> <( )( )( )( <( )( Sunny <( <( <( Working )> )( )> )> or Not working or Rainy )> is NOT frequent but: • All not working and sunny • Does NOT appear elsewhere Problem: too small proportion of not working and sunny

Pattern Discovery: What if contextual information is available? Rosa Frédérick François Bruno Pascal Osmar Sandra Maguelonne Julien Alberto Contextual Information <( )( )> )( )> <( )( )( )( <( )( Sunny <( <( <( Working or Not working or Rainy )> is frequent: • For all day types )> )( )> )> • Considered as generally frequent