Forum Stats

  • 3,826,346 Users
  • 2,260,631 Discussions


need advices... pls help

3027851 Member Posts: 10
edited Sep 30, 2015 9:30PM in Machine Learning

guys, basically i am very new to data mining.

i have a client who has been emphasizing on "patterns", "patterns" and "patterns".

Basically, the project im involved in is in the aviation industry and the objective is to find out if there are any "patterns" of the Air Traffic Controller while controlling the aircraft.

All i had for the data includes the following information:

1. Controller Position - [VARCHAR]

2. Sector (the sector that the aircraft is in during the control) - [VARCHAR]

3. Controller inputs (the inputs used to communicate with the pilot) - [VARCHAR]

4. Flight ID (unique ID of the aircraft) - [NUMERIC]

5. Controller input time (time of controller input) - [DATETIME]

Above data are stored in a Transactional format. i.e. 1 aircraft can have multiple rows of controller inputs issued to it.

Now, correct me if I am wrong. For classification models, there need to have an predefined class variable right? Classification models are more for predictions rather than sifting out patterns?

Next, looking at the type of information I have, it is not suitable for Clustering ? as in all the data are in [VARCHAR] ?

If classification and clustering is not applicable, then I am left with association. however, ODM 11g does not support sequence pattern mining...

I am at a loss now.. any experts please advise what should i do..... R is not possible due to certain reasons.


This discussion has been closed.