Submitted by: Submitted by PureMuscle12
Views: 273
Words: 274
Pages: 2
Category: Business and Industry
Date Submitted: 05/04/2012 12:27 PM
Assignment 4
Classification Tree
Decision Rules
If Vintage = 2000, 2001, 2002, 2003 and the Style of Wine is Blend then classify the wine as getting a rating of 3 stars (Th)
If Vintage = 2000, 2001, 2002, 2003 and the Style of Wine is Carbernet Savignon, Shiraz or Pinot Noir, then classify the rating of the wine as getting 3.5 stars (ThH)
If Vintage = 2004, 2005, 2006, 2007, 2008, 2009 (Possibly 2004-2009?) and the Style of wine is Blend, then classify the rating of the wine as getting 3.5 stars (ThH)
If Vintage = 2004, 2005, 2006, 2007, 2008, 2009 and the Style of wine is Cabernet Savignon, Shiraz or Pinot Noir, then classify the rating of the wine as 4.5 stars (FoH).
Predicted Class x Observed Class n's (Assignment4_Winedata) Predicted (row) x observed (column) matrix Learning sample N = 1500 |
| Class - ThH | Class - Fo | Class - Fi | Class - Th | Class - FoH |
ThH | 336 | 248 | 14 | 14 | 45 |
Fo | 0 | 0 | 0 | 0 | 0 |
Fi | 0 | 0 | 0 | 0 | 0 |
Th | 66 | 0 | 0 | 74 | 0 |
FoH | 19 | 188 | 101 | 0 | 395 |
Hit Rate (Training Sample): 336+74+3951500 = 53.67%
Test Sample Misclassification Matrix (Assignment4_Winedata) Predicted (row) x observed (column) matrix CV cost = .448; s.d. CV cost = .02224 |
| Class - ThH | Class - Fo | Class - Fi | Class - Th | Class - FoH |
ThH | | 80 | 5 | 7 | 15 |
Fo | 0 | | 0 | 0 | 0 |
Fi | 0 | 0 | | 0 | 0 |
Th | 14 | 0 | 0 | | 0 |
FoH | 6 | 68 | 29 | 0 | |
Hit Rate (Test Sample):
500-(80+5+7+15+14+6+68+29500 = 55.2% (Potentially worrying as I got a higher hit-rate for Test Sample than Training sample)