def __init__(self):
datarows, labels = semdatautil.get_sem_data('semeion.data')
# SPLIITING IN HALF WILL WORK SINCE IT IS NOT JUST 0 1 2 3 4 5 6 7 8 9 IT IS 0 1 ... 9 0 1 2 ... 9 0 1 2 ETC
self.trainingdatarows = datarows[len(datarows)/2:]
self.traininglabels = labels[len(labels)/2:]
self.testingdatarows = datarows[:len(datarows)/2]
self.testinglabels = labels[:len(labels)/2:]
self.trainingpymldata = PyML.VectorDataSet(self.trainingdatarows, L=self.traininglabels)
self.testingpymldata = PyML.VectorDataSet(self.testingdatarows, L=self.testinglabels)
rf = open('results', 'w')
rf.write("========== RESULTS ==========\n")
rf.close()
super(DataViewer, self).run_tk_mainloop()
def show_data(self, datano):
self.currdatano = datano
self.ax.imshow(semdatautil.sem_datarow_to_image(self.datarows[datano]), cmap=pyplot.gray())
self.fig.canvas.draw()
print self.currdatano
if self.labels != None:
print "Label: " + str(self.labels[self.currdatano])
def next_data(self):
self.show_data(self.currdatano + 1)
def prev_data(self):
if self.currdatano > 0:
self.show_data(self.currdatano - 1)
if __name__ == "__main__":
fig = Figure() # figsize=(10,10))#, dpi=100)
ax = fig.add_subplot(111)
ax.set_title("Data")
datarows, labels = semdatautil.get_sem_data("semeion.data")
print labels
print datarows[0]
dataviewer = DataViewer(fig, ax, datarows, labels=labels)
return multiclass_predicted_labels
def prediction_accuracy(predicted_labels, actual_labels):
numcorrect = 0
for i in range(len(predicted_labels)):
if int(predicted_labels[i]) == int(actual_labels[i]):
numcorrect += 1
return numcorrect, len(predicted_labels), float(numcorrect)/float(len(predicted_labels))
if __name__ == "__main__":
rf = open('results', 'w')
valid_labels = [0,1,2,3,4,5,6,7,8,9]
datarows, o_datalabels = semdatautil.get_sem_data('semeion.data')
print len(datarows)
# convert the data to numeric labels
datalabels = [int(label) for label in o_datalabels]
# SPLIITING IN HALF WILL WORK SINCE IT IS NOT JUST 0 1 2 3 4 5 6 7 8 9 IT IS 0 1 ... 9 0 1 2 ... 9 0 1 2 ETC
trainingdata = datarows[:len(datarows)/3]
traininglabels = datalabels[:len(datalabels)/3]
paramseldata = datarows[len(datarows)/3:(len(datarows)/3)*2]
paramsellabels = datalabels[len(datalabels)/3:(len(datalabels)/3)*2]
testingdata = datarows[(len(datarows)/3)*2:]
