def kmeans(data, **kwargs):
"""
Perform k-means clustering on unstructured N-dimensional data.
@type data: array
@param data: The data to be clustered
@type kwargs: dict
@param kwargs: The following args are accepted:
- numClusters: The number of clusters to form (returned number of clusters may be less than k).
- npasses: The number of times the k-means clustering algorithm is performed,
each time with a different (random) initial condition.
- method: describes how the center of a cluster is found:
- method=='a': arithmetic mean.
- method=='m': median.
- initialCenters: a set of points that should be used as the initial
cluster centers
@rtype: tuple
@return: A list where each element indicates the cluster membership of the
corresponding index in the original data and a message string
"""
k = 1
npasses = 1
method = 'a'
initialCenters = None
smartCenters = False
msg = ''
if 'numClusters' in kwargs:
k = int(kwargs['numClusters'])
if 'npasses' in kwargs:
npasses = int(kwargs['npasses'])
if 'method' in kwargs:
method = kwargs['method']
if 'initialCenters' in kwargs:
initialCenters = kwargs['initialCenters']
if 'smartCenters' in kwargs:
smartCenters = kwargs['smartCenters']
logData = tm.getMethod('log')(data)
if initialCenters is not None:
(clusterIDs, err, nOpt) = pc.kcluster(logData, k, npass=npasses, method=method)
msg = "Number of rounds optimal solution was found: %i" % nOpt
else:
logCenters = tm.getMethod('log')(np.array(initialCenters[:k]))
(centroids, clusterIDs) = kmeans2(logData, logCenters, minit='matrix')
if len(np.unique(clusterIDs)) < k:
wx.MessageBox('Warning: One or more of the returned clusters are empty. Please choose different initial cluster centers and re-run k-means for better results.', 'Insufficiently varied cluster centers', wx.OK | wx.ICON_WARNING)
return clusterIDs, msg
def heatmap2d(subplot, figure, dims):
"""
heatmap2d; 2D Heatmap; Plots a 2D heatmap of the data with event density as the main indicator.
"""
opts = subplot.opts
if len(opts) == 0:
opts['type'] = 'Hexbins'
opts['colorMap'] = 'gist_earth'
opts['bins'] = (200, 200)
opts['transform'] = 'log'
opts['transformAuto'] = True
subplot.axes = figure.add_subplot(subplot.mnp, title=subplot.Title)
subplot.axes.set_xlabel(subplot.Labels[dims[0]])
subplot.axes.set_ylabel(subplot.Labels[dims[1]])
x = subplot.Data[:, dims[0]]
y = subplot.Data[:, dims[1]]
cbLabel = ''
# apply transform
if 'transform' not in opts:
opts['transform'] = 'log'
if opts['transform'] == 'log':
x = tm.getMethod('log')(x)
y = tm.getMethod('log')(y)
extent = (0, x.max()*1.05, 0, y.max()*1.05)
cmap = CM.get_cmap(opts['colorMap'])
if opts['type'] == 'Hexbins':
gAx = subplot.axes.hexbin(x, y, gridsize=opts['bins'][0], extent=extent, mincnt=1, cmap=cmap)
cbLabel = 'Events'
# The following two means of calculating the heat map do not
# work correctly yet and are not enabled for selection.
if opts['type'] == 'Gaussian KDE':
kdeGrid = fast_kde(x, y, gridsize=opts['bins'])
gAx = subplot.axes.imshow(kdeGrid, extent=extent, cmap=cmap,
aspect='auto', interpolation='bicubic')
if opts['type'] == 'Histogram':
heatmap, xedges, yedges = np.histogram2d(x, y, bins=opts['bins'])
extent = [xedges[0], xedges[-1], yedges[0], yedges[-1]]
gAx = subplot.axes.imshow(heatmap, extent=extent, cmap=cmap,
aspect='auto')
cb = subplot.parent.colorbar(gAx)
if cbLabel == '':
cbLabel = 'Events' if opts['transform'] != 'log' else 'log Events'
cb.set_label(cbLabel)
def histogram(subplot, figure, dims):
"""
histogram; Histogram; Plots a 1D Histogram
"""
# set default plot options if necessary
opts = subplot.opts
if len(opts) == 0:
opts['type'] = 'Gaussian KDE'
opts['bins'] = 200
opts['transformAuto'] = True
opts['xTransform'] = ''
opts['yTransform'] = ''
opts['kdeDisplay'] = True
# Set axes transforms
if (opts['transformAuto']):
opts['xTransform'] = 'log'
opts['yTransform'] = 'linear'
subplot.axes = figure.add_subplot(subplot.mnp, title=subplot.Title)
subplot.axes.set_xlabel(subplot.Labels[dims[0]])
data = subplot.Data[:, dims[0]]
if opts['xTransform'] == 'log':
data = tm.getMethod('log')(data)
# Kernel density estimation
if opts['type'] == 'Gaussian KDE' or opts['type'] == 'Both':
ind = np.linspace(np.min(data), np.max(data), data.shape[0]*.1)
gkde = stats.gaussian_kde(data)
kdepdf = gkde.evaluate(ind)
subplot.axes.plot(ind, kdepdf, label='kde', color='blue')
# Binned Histogram
if opts['type'] != 'Gaussian KDE':
#subplot.axes.hist(subplot.Data[:, dims[0]], bins=250, normed=True, histtype='bar',log=True)
h, b = np.histogram(data, bins=opts['bins'])
if opts['type'] == 'Both':
h = tm.getMethod('log')(h)
b = (b[:-1] + b[1:])/2.0
subplot.axes.plot(b, h)
if opts['type'] == 'Both':
dataMax = max(np.max(kdepdf), np.max(h))
subplot.axes.set_ylim(0, dataMax + 0.1)
def bakker_kMeans(data, **kwargs):
"""
This is an implementation of the k-means algorithm designed specifically
for flow cytometry data in the following paper:
T.C.B. Schut, B.G.D. Grooth, and J. Greve,
"Cluster analysis of flow cytometric list mode data on a personal computer",
Cytometry, vol. 14, 1993, pp. 649-659.
@type data: array
@param data: The data to be clustered
@type kwargs: dict
@param kwargs: The following args are accepted:
- numClusters: The number of clusters to form
@rtype: tuple
@return: A list where each element indicates the cluster membership of the
corresponding index in the original data and a message string
"""
k = 1
initClusters = 200
msg = ''
if 'numClusters' in kwargs.keys():
k = int(kwargs['numClusters'])
if 'initClusters' in kwargs.keys():
initClusters = int(kwargs['numClusters'])
# Log transform
logData = tm.getMethod('log')(data)
# Choose large # (200 as suggested by authors) of non-random initial centers
centers = util.kinit(logData, initClusters)
# Run k-means
_, ids = kmeans2(logData, np.array(centers), minit='matrix')
# Merge clusters w/special comparison metric until user cluster # achieved
clusters = util.separate(logData, ids)
finalIDs = merge(k, ids, clusters)
return finalIDs, msg
