| | |
- JumpWithinKMedoids_cluster(distanceMatrix, kmax=10, num_cycles=5, max_kMedoids_iterations=1000, min_dist=0.14999999999999999, seeds=None, verbose=0, print_fnctn=None, data=None)
- (There are better choices) Increases k over a range (1-kmax) and selects k for which
the change from k-1 "within_distance" is largest.
- MinWithinKMedoids_cluster(distanceMatrix, k, num_cycles=5, max_kMedoids_iterations=1000, seeds=None, verbose=0, print_fnctn=None, data=None, return_WD=None)
- Like original KMedoids_cluster, but the optimal iteration is picked based on
having the minimum 'within distance', which is computed by the withinDist
function.
WD = SUM_clusters (SUM_motifs (SUM_inter_motif_distances)))
There is an optional flag to return WD to the parent, in case it wants to use it
decide whether to repeat with a different value of k, as in "JumpWithin"
- averageList(lst)
- bestKMedoids_cluster(distanceMatrix, kmax=10, num_cycles=5, max_kMedoids_iterations=1000, min_dist=0.14999999999999999, seeds=None, verbose=0, print_fnctn=None, data=None)
- Descending k: Stop Descending when distances between medoids becomes
larger than min_dist
- bestaveKMedoids_cluster(distanceMatrix, kmax=10, num_cycles=5, max_kMedoids_iterations=1000, min_dist=0.14999999999999999, seeds=None, verbose=0, print_fnctn=None, data=None, kmin=0)
- [Ben's Current pick 12-02-03]
Descending: Compute ave dist of members of a cluster to each of the other medoids.
If any average distances are too small (<min_dist), reduce k
- bestupKMedoids_cluster(distanceMatrix, kmax=10, num_cycles=5, max_kMedoids_iterations=1000, min_dist=0.14999999999999999, seeds=None, verbose=0, print_fnctn=None, data=None)
- Ascending (there are better choices)
Ascending version of "best", which looks for the point at which the
distances between medoids becomes too small
- generate_random_1d(x=0, mult=1.0, n=100)
- generate_random_2d(x=0, y=0, mult=1.0, n=100)
- kMedoids_cluster(distanceMatrix, k, max_iterations=1000, seed='12345')
- use kMedoid clustering
return a three-tuple:
1. k indices representing the medoids
2. k-lists of indices representing the clusters,
3. the average distance to the medoid in each cluster
- metaKMedoids_cluster(distanceMatrix, k, num_cycles=5, max_kMedoids_iterations=1000, seeds=None, verbose=0, print_fnctn=None, data=None)
- ###NOTE CHANGED FROM Fraenkel.Clustering###
run kMedoid clustering num_cycles times
return a three-tuple:
1. k indices representing the medoids
2. k-lists of indices representing the clusters,
3. the average distance to the medoid in each cluster
if verbose =1 and a print function, print_fnctn, and the original data (not just the distance matrix) are supplied
the program will print out intermediate clustering results
- metaminDiamKMedoids_cluster(distanceMatrix, kstart=1, num_cycles=5, max_kMedoids_iterations=1000, min_dist=0.14999999999999999, seeds=None, verbose=0, print_fnctn=None, data=None)
- Ascending: increase k until the Diam (a.k.a. "distances") for each cluster are
smaller than min_dist
- minDiamKMedoids_cluster(distanceMatrix, k, num_cycles=5, max_kMedoids_iterations=1000, seeds=None, verbose=0, print_fnctn=None, data=None)
- print_Dmat(D)
- print_list_clustering_results(clusters, data, avg_dist, k)
- print_motif_clustering_results(clusters, data, avg_dist, k)
- withinDist(clusters, data)
|