Assignment 1
Note:
This assignment should be done by each student individually.
You can discuss it in general terms with other students; however,
the files you hand in, e.g., the report and codes should be your
own. If I find your reports/codes are the same as or similar to
the others, both of you cannot get the scores for this assignment.
1. Problem Statement
In this assignment, you have one data set with 3000 samples. You can
randomly select 2500 samples as training data, and 500 samples as test data.
Then, you are required to classify the test data using Bayesian Theorem. To
estimate the densities, you need to use one parameter estimation (e.g., MLE
or Bayesian estimation) and one non-parameter estimation method (e.g.,
KNN or Parzen window), respectively (Select one you like). For each model,
you should give the detail that you have to consider, e.g., how do you select
the parameters in the model. Finally, you should list and analyze the results
of each model including the test errors and the impact of the parameters,
etc.
By Xin-Shun Xu November 14, 2013
- 1
- 2
- 3
- 4
- 5
- 6
前往页