SUMMARY

• LDA and DT classification techniques were studied and applied to the phone classification problem

• These two techniques were also applied for a simpler case to help visualize the classification criterion

• DT's seem to do very well on closed loop testing but fall short on open loop testing due to over-training

• The challenge lies in defining an appropriate stopping rule for the DT's during training - presently they train till all data elements ae classified properly, leading to over-training