Transcript Vehicle

Vehicle
• We had to design a data mining tool, to identify
4 different types of vehicles based on their
silhouettes.
• This could be used for a toll gate, which
charges different prices for different types of
vehicles.
Frank and Nick
The problem and the goal
To classify a given silhouette as one of four
types of vehicle, using a set of features
extracted from the silhouette. The vehicle
may be viewed from one of many different
angles.
Description of data
• There are 19 attributes in the data set
• There are 846 instances in the data set
Description of data
• Some of the attributes include:
-
Compactness
Circularity
Scatter ratio
Scaled radius of gyration
Radius ratio
Etc.
Classification accuracy
• Decision stump: 38.5%
• Decision tree: 69.7%
• ZeroR: 21.5%
Interesting patterns
• If max length aspect ratio is between 7.3
and infinity then it’s a bus
• If pr. Axis rectangularity is between 25.4
and infinity then it’s a bus
Association rules
• If pr. axis rectangularity is between 18.2
and 19.4 then scatter ratio is between
142.6 and 157.9
• If elongatedness is between 43.5 and 47
and pr. axis rectangularity is between 18.2
and 19.4 then scatter ratio is between
142.6 and 157.9
Practical use
• The results could be used to determine the
price that a vehicle has to pay at a toll
gate, depending on what type of vehicle it
is.