def predict():
"""
Method: POST
Predict topic ids from given docs
:return: topic ids
"""
content = request.json
if not isinstance(content, dict):
return 'Not json format', 400
if "docs" not in content:
return 'field `docs` not present', 400
docs = content['docs']
if not isinstance(docs, list) or len(docs) == 0:
return '`docs` should be list and not empty', 400
for doc in docs:
if not isinstance(doc, str):
return 'Wrong data type of doc', 400
topics = kmeans.predict(docs).tolist()
return {'topics': topics}, 200
def process(selectedCrime, type):
colors = [
'red', 'orange', 'yellow', 'green', 'blue', 'purple', 'black', 'white',
'cyan', 'brown'
]
finalData = {}
for year in range(2005, 2017):
print(year)
finalData[str(year)] = {}
for month in range(1, 13):
print(' ' + str(month))
finalData[str(year)][str(month)] = {}
for num_clusters in range(2, 11):
print(' ' + str(num_clusters))
points = {"type": "FeatureCollection", "features": []}
finalData[str(year)][str(month)][str(num_clusters)] = {}
clusterData = [[] for Null in range(num_clusters)]
geodesic_clusters, coordinate_array, data_array = get_clusters(
selectedCrime=selectedCrime,
type='geodesic',
month=month,
year=year,
num_clusters=num_clusters)
euclidean_clusters, co, da = get_clusters(
selectedCrime=selectedCrime,
type='euclidean',
month=month,
year=year,
num_clusters=num_clusters)
for coor in coordinate_array:
c_geodesic = predict(
num_clusters,
geodesic_clusters,
[float(coor[0]), float(coor[1])],
type='geodesic')
c_euclidean = predict(
num_clusters,
euclidean_clusters,
[float(coor[0]), float(coor[1])],
type='euclidean')
clusterData[int(c_geodesic)].append(
[float(coor[1]), float(coor[0])])
points["features"].append({
"geometry": {
"type": "Point",
"coordinates": [float(coor[1]),
float(coor[0])]
},
"type": "Feature",
"properties": {
"fillColor": colors[c_euclidean],
"euclidean_cluster": c_euclidean,
"geodesic_cluster": c_geodesic,
"popupContent": ""
},
})
finalData[str(year)][str(month)][str(
num_clusters)]['points'] = points
for i in range(0, num_clusters):
try:
finalData[str(year)][str(month)][str(num_clusters)][
str(i)] = build_cluster_data(
clusterData[i],
selected_crime=selectedCrime,
cluster=i)
except:
try:
finalData[str(year)][str(month)][str(
num_clusters)][str(i)] = build_cluster_data(
clusterData[i],
selected_crime=selectedCrime,
cluster=i)
except:
print('continuing')
continue
fn = open('../static/' + selectedCrime + '/' + type + '_final_points.js',
'w')
fn.write('var geodesic_data = ')
fn.write(json.dumps(finalData))
fn.write(';\n')
fn.close()
from sklearn.datasets import load_digits
from sklearn.metrics import fowlkes_mallows_score
from sklearn.cluster import AgglomerativeClustering, AffinityPropagation
from kmeans import kmeans, Point, predict
data, target = load_digits(return_X_y=True)
# K-Means
kmeans_data = [Point(val) for val in data]
k_means = kmeans(kmeans_data, 10)
labels = []
for point in data:
labels.append(predict(k_means, Point(point)))
target = [int(num) for num in target]
results = [[0 for _ in range(10)] for __ in range(10)]
for i, val in enumerate(labels):
results[target[i]][val] += 1
conversion = {}
for t_i, targ in enumerate(results):
max_cluster = None
for c_i, cluster in enumerate(targ):
if max_cluster is None or cluster > targ[max_cluster]:
max_cluster = c_i
conversion[t_i] = max_cluster
print "source data's length: ", len(tranSourceFeature)
#experiment 1
testData, testLabel = readTestData(targetPath)
print "target outlier length: ", len(testLabel)
ori_testData = testData
K = 5
transferMatrix = np.array(transferMatrix).T
testData = np.array(testData).T
outlierTest = dot(transferMatrix, testData)
outlierTest = outlierTest.T
outlierTest.tolist()
sigma = 0.0005
kernelTranSourceFeature = kernel(tranSourceFeature, sigma)
centroids, clusterAssment, radiusCluster = kmeansAlgorithm(
mat(kernelTranSourceFeature), K)
numSamples = len(data)
# clusterAssment = mat(zeros((numSamples, 1)))
# for i in xrange(numSamples):
# clusterAssment[i, :] = predict_label[i]
purity = calcPurity(clusterAssment, K, sourceLabel)
print "purity", purity
sumPurity = 0.0
for i in purity:
sumPurity += i
print "avg purity", sumPurity / len(purity)
predict(mat(kernel(tranTargetFeature, sigma)), centroids, K, radiusCluster)
# Pulses, E, PVI
prefix = os.environ["INFUSION_DATASETS"]
data = torch.load(f'{prefix}/pulses_E_PVI.pt')
# Clustering
km = torch.load("_kmeans/km.pt")
# Model
st = torch.load('st/out64/mon3')
model.load_state_dict(st)
model.train(False)
#--- Distribution of E within cluster ---
x = data["pulses"].view([-1, 128])
y = model(x)
c = kmeans.predict(km, y)
E = (data["E"].unsqueeze(1).repeat(1, 256).view([-1]))
df = pd.DataFrame({"E": E.numpy(), "C": c.numpy()})
pivot = df.pivot(columns="C", values="E")
def seriesE(i):
return pivot[i][pivot[i].notnull()]
ax = sns.catplot(y="E",
col="C",
col_wrap=8,
kind="violin",