def main():
# Create 2 artificial clusters that partially overlap
X,y = Tomeklink.createCluster()
print y
# Plot the clusters
colors = np.array([x for x in 'bgrcmykbgrcmykbgrcmykbgrcmyk'])
colors = np.hstack([colors] * 20)
pl.scatter(np.array(X)[:, 0], np.array(X)[:, 1], color=colors[y].tolist(), s=10)
pl.show()
# Detect the TomekLinks in the data
tomeklinks = Tomeklink.detectTomekLinks(X,y)
# Remove the TomekLinks from the data
X,y = Tomeklink.removeTomekLinks(tomeklinks,X,y)
# Plat the data again with the TomekLinks removed
colors = np.array([x for x in 'bgrcmykbgrcmykbgrcmykbgrcmyk'])
colors = np.hstack([colors] * 20)
pl.scatter(np.array(X)[:, 0], np.array(X)[:, 1], color=colors[y].tolist(), s=10)
pl.show()
unhealthyMatrix = data[data[0] > 0].as_matrix()
healthyMatrix = data[data[0] < 0].as_matrix()
oversample = smote.SMOTE(np.array(unhealthyMatrix), 1000, 5)
# final= healthy + unhealthyMatrix oversampling
final = np.concatenate((healthyMatrix, oversample), axis=0)
# svm
X = [sub[1:] for sub in final]
Y = []
for sub in final:
Y.append(sub[0])
# Detect the TomekLinks in the data
tomeklinks = Tomeklink.detectTomekLinks(X, Y)
# Remove the TomekLinks from the data
X, Y = Tomeklink.removeTomekLinks(tomeklinks, X, Y)
clf = svm.SVC()
clf.fit(X, Y)
# predict
test = []
data = pd.read_table(path2, header=None)
for row in data.iterrows():
test.append(row[1].tolist())
X_test = [sub[1:] for sub in test]
rightAns = []
