This preview shows page 1. Sign up to view the full content.
Unformatted text preview: ore the overall accuracies you obtain for each value of n in a vector OA_original
(a 1x99 vector).
b) In the previous case, you did not perform any dimensionality reduction. Now, repeat the
experiment in (a) except that before you perform classification, apply a Linear Discriminant
Analysis dimensionality reduction transform to f1Train, f2Train and fTest. Train and test your
classifier in the reduced dimensional space after you have applied LDA to all the data. Store the
overall accuracies for each value of n in a vector OA_LDA (a 1x99 vector). Note that when you
apply the LDA transformation in this problem, you are supposed to learn your LDA
transformation using only the training data (f1Train and f2Train), and apply this same
transformation on both the training and test dataset.
c) On the same plot, show overall accuracies with and without LDA as a function of n (n on the xaxis, Overall Accuracy on the y-axis). What do you conclude from this plot? Explain in detail.
10 points 4 of 4 Instructor: Dr. Saurabh Prasad Email: firstname.lastname@example.org Due by 10am, February 26, 2010 ...
View Full Document
- Spring '10