Algorithm Settings for Non-Negative Matrix Factorization
Non-Negative Matrix Factorization (NMF), the algorithm for feature extraction, has the following settings:
- Minimum Convergence Tolerance: The number indicating minimum convergence tolerance value; default is 0.5.
- Maximum Iterations: The integer indicating the maximum number of iterations to be performed; the default is 50.
- Do you want to specify the number of Features?: The default is No, in which case the algorithm determines the number of features.
If you wish to specify the number of features, click the radio button beside Yes, and type in the integer number of features. The number of features must be a positive integer less than or equal to the minimum of the number of attributes and the number of cases; in many cases, 5 or some other number less than or equal to 7 will give good results.
- Specify Random Seed?: Specify a random seed for the sample. The default is Yes; 9999 is the default value of the seed. The seed can be changed; the value of the seed must be an integer > 0. If you plan to repeat this operation and to get the same results, make sure to use the same random seed. You can also select No and not provide a random seed.
Copyright © 2006, 2008, Oracle. All rights
reserved.