gemseo / third_party

sompy module

Self Organizing Maps

Classes:

SOM(name, Data[, mapsize, norm_method, ...])

name and data, neigh== Bubble or Guassian

Functions:

batchtrain(self[, njob, phase, ...])	param njob Default value = 1)
chunk_based_bmu_find(self, x, y, y_2)	param x param y:
denormalize_by(data_by, n_vect[, n_method])	param data_by param n_vect:
grid_dist(self, bmu_ind)	som and bmu_ind depending on the lattice "hexa" or "rect" we have different grid distance functions.
l(a, b)	param a param b:
lininit(self)
normalize(data[, method])	param data param method: (Default value = 'var')
normalize_by(data_raw, data[, method])	param data_raw param data:
rect_dist(self, bmu)	param bmu
view_1d(self, text_size[, which_dim, what])	param text_size param which_dim: (Default value = 'all')
view_2d(self, text_size[, which_dim, what])	param text_size param which_dim: (Default value = 'all')
view_2d_Pack(self, text_size[, which_dim, ...])	param text_size param which_dim: (Default value = 'all')

class gemseo.third_party.sompy.SOM(name, Data, mapsize=None, norm_method='var', initmethod='pca', neigh='Guassian')

Bases: object

name and data, neigh== Bubble or Guassian

Methods:

calc_map_dist()
cluster([method, n_clusters])	param method Default value = 'Kmeans')
find_K_nodes(data[, K])	param data param K: (Default value = 5)
hit_map([data])	param data Default value = None)
hit_map_cluster_number([data])	param data Default value = None)
ind_to_xy(bm_ind)	param bm_ind
init_map()
node_Activation(data[, wt, Target])	‘uniform’
para_bmu_find(x, y[, njb])	param x param y:
predict(X_test[, K, wt])	‘uniform’
predict_Probability(data, Target[, K])	param data param Target:
predict_by(data, Target[, K, wt])	‘uniform’
project_data(data)	param data
set_algorithm([initmethod, algtype, ...])	initmethod = ['random', 'pca'] algos = ['seq','batch'] all_neigh = ['gaussian','manhatan','bubble','cut_gaussian','epanechicov' ] alfa_types = ['linear','inv','power']
set_data_labels([dlabel])	param dlabel Default value = None)
set_topology([mapsize, mapshape, lattice, ...])	all_mapshapes = ['planar','toroid','cylinder'] all_lattices = ['hexa','rect']
train([trainlen, n_job, shared_memory, verbose])	param trainlen Default value = None)
update_codebook_voronoi(training_data, bmu, ...)	param training_data param bmu:
view_map([what, which_dim, pack, text_size, ...])	param what Default value = 'codebook')

calc_map_dist()

cluster(method='Kmeans', n_clusters=8)

Parameters

• method –

  Default value = ‘Kmeans’)

  By default it is set to Kmeans.

• n_clusters –

  Default value = 8)

  By default it is set to 8.

find_K_nodes(data, K=5)

Parameters

• data – param K: (Default value = 5)

• K –

  (Default value = 5)

  By default it is set to 5.

hit_map(data=None)

Parameters

data –

Default value = None)

By default it is set to None.

hit_map_cluster_number(data=None)

Parameters

data –

Default value = None)

By default it is set to None.

ind_to_xy(bm_ind)

Parameters

bm_ind –

init_map()

node_Activation(data, wt='distance', Target=None)

‘uniform’

Parameters

• data – param wt: (Default value = ‘distance’)

• Target –

  Default value = None)

  By default it is set to None.

• wt –

  (Default value = ‘distance’)

  By default it is set to distance.

para_bmu_find(x, y, njb=1)

Parameters

• x – param y:

• njb –

  Default value = 1)

  By default it is set to 1.

• y –

predict(X_test, K=5, wt='distance')

‘uniform’

Parameters

• X_test – param K: (Default value = 5)

• wt –

  Default value = ‘distance’)

  By default it is set to distance.

• K –

  (Default value = 5)

  By default it is set to 5.

predict_Probability(data, Target, K=5)

Parameters

• data – param Target:

• K –

  Default value = 5)

  By default it is set to 5.

• Target –

predict_by(data, Target, K=5, wt='distance')

‘uniform’

Parameters

• data – param Target:

• K –

  Default value = 5)

  By default it is set to 5.

• wt –

  Default value = ‘distance’)

  By default it is set to distance.

• Target –

project_data(data)

Parameters

data –

set_algorithm(initmethod='pca', algtype='batch', neighborhoodmethod='gaussian', alfatype='inv', alfaini=0.5, alfafinal=0.005)

initmethod = [‘random’, ‘pca’] algos = [‘seq’,’batch’] all_neigh = [‘gaussian’,’manhatan’,’bubble’,’cut_gaussian’,’epanechicov’ ] alfa_types = [‘linear’,’inv’,’power’]

Parameters

• initmethod –

  Default value = ‘pca’)

  By default it is set to pca.

• algtype –

  Default value = ‘batch’)

  By default it is set to batch.

• neighborhoodmethod –

  Default value = ‘gaussian’)

  By default it is set to gaussian.

• alfatype –

  Default value = ‘inv’)

  By default it is set to inv.

• alfaini –

  Default value = .5)

  By default it is set to 0.5.

• alfafinal –

  Default value = .005)

  By default it is set to 0.005.

set_data_labels(dlabel=None)

Parameters

dlabel –

Default value = None)

By default it is set to None.

set_topology(mapsize=None, mapshape='planar', lattice='rect', mask=None, compname=None)

all_mapshapes = [‘planar’,’toroid’,’cylinder’] all_lattices = [‘hexa’,’rect’]

Parameters

• mapsize –

  Default value = None)

  By default it is set to None.

• mapshape –

  Default value = ‘planar’)

  By default it is set to planar.

• lattice –

  Default value = ‘rect’)

  By default it is set to rect.

• mask –

  Default value = None)

  By default it is set to None.

• compname –

  Default value = None)

  By default it is set to None.

train(trainlen=None, n_job=1, shared_memory='no', verbose='on')

Parameters

• trainlen –

  Default value = None)

  By default it is set to None.

• n_job –

  Default value = 1)

  By default it is set to 1.

• shared_memory –

  Default value = ‘no’)

  By default it is set to no.

• verbose –

  Default value = ‘on’)

  By default it is set to on.

update_codebook_voronoi(training_data, bmu, H, radius)

Parameters

• training_data – param bmu:

• H – param radius:

• bmu –

• radius –

view_map(what='codebook', which_dim='all', pack='Yes', text_size=2.8, save='No', save_dir='empty', grid='No', text='Yes', cmap='None', COL_SiZe=6)

Parameters

• what –

  Default value = ‘codebook’)

  By default it is set to codebook.

• which_dim –

  Default value = ‘all’)

  By default it is set to all.

• pack –

  Default value = ‘Yes’)

  By default it is set to Yes.

• text_size –

  Default value = 2.8)

  By default it is set to 2.8.

• save –

  Default value = ‘No’)

  By default it is set to No.

• save_dir –

  Default value = ‘empty’)

  By default it is set to empty.

• grid –

  Default value = ‘No’)

  By default it is set to No.

• text –

  Default value = ‘Yes’)

  By default it is set to Yes.

• cmap –

  Default value = ‘None’)

  By default it is set to None.

• COL_SiZe –

  Default value = 6)

  By default it is set to 6.

gemseo.third_party.sompy.batchtrain(self, njob=1, phase=None, shared_memory='no', verbose='on')

Parameters

• njob –

  Default value = 1)

  By default it is set to 1.

• phase –

  Default value = None)

  By default it is set to None.

• shared_memory –

  Default value = ‘no’)

  By default it is set to no.

• verbose –

  Default value = ‘on’)

  By default it is set to on.

gemseo.third_party.sompy.chunk_based_bmu_find(self, x, y, y_2)

Parameters

• x – param y:

• y_2 –

• y –

gemseo.third_party.sompy.denormalize_by(data_by, n_vect, n_method='var')

Parameters

• data_by – param n_vect:

• n_method –

  Default value = ‘var’)

  By default it is set to var.

• n_vect –

gemseo.third_party.sompy.grid_dist(self, bmu_ind)

som and bmu_ind depending on the lattice “hexa” or “rect” we have different grid distance functions. bmu_ind is a number between 0 and number of nodes-1. depending on the map size bmu_coord will be calculated and then distance matrix in the map will be returned

Parameters

bmu_ind –

gemseo.third_party.sompy.l(a, b)

Parameters

• a – param b:

• b –

gemseo.third_party.sompy.lininit(self)

gemseo.third_party.sompy.normalize(data, method='var')

Parameters

• data – param method: (Default value = ‘var’)

• method –

  (Default value = ‘var’)

  By default it is set to var.

gemseo.third_party.sompy.normalize_by(data_raw, data, method='var')

Parameters

• data_raw – param data:

• method –

  Default value = ‘var’)

  By default it is set to var.

• data –

gemseo.third_party.sompy.rect_dist(self, bmu)

Parameters

bmu –

gemseo.third_party.sompy.view_1d(self, text_size, which_dim='all', what='codebook')

Parameters

• text_size – param which_dim: (Default value = ‘all’)

• what –

  Default value = ‘codebook’)

  By default it is set to codebook.

• which_dim –

  (Default value = ‘all’)

  By default it is set to all.

gemseo.third_party.sompy.view_2d(self, text_size, which_dim='all', what='codebook')

Parameters

• text_size – param which_dim: (Default value = ‘all’)

• what –

  Default value = ‘codebook’)

  By default it is set to codebook.

• which_dim –

  (Default value = ‘all’)

  By default it is set to all.

gemseo.third_party.sompy.view_2d_Pack(self, text_size, which_dim='all', what='codebook', save='No', grid='Yes', save_dir='empty', text='Yes', CMAP='None', col_sz=None)

Parameters

• text_size – param which_dim: (Default value = ‘all’)

• what –

  Default value = ‘codebook’)

  By default it is set to codebook.

• save –

  Default value = ‘No’)

  By default it is set to No.

• grid –

  Default value = ‘Yes’)

  By default it is set to Yes.

• save_dir –

  Default value = ‘empty’)

  By default it is set to empty.

• text –

  Default value = ‘Yes’)

  By default it is set to Yes.

• CMAP –

  Default value = ‘None’)

  By default it is set to None.

• col_sz –

  Default value = None)

  By default it is set to None.

• which_dim –

  (Default value = ‘all’)

  By default it is set to all.