def findNearestCorrects(self, assn, asts, numASTs, label):
nn = {}
if label == 'corrects':
for ast in asts:
nn[ast] = ast
return nn
corrects = self.getAsts(assn, 'corrects')
sources = asts[:numASTs]
distanceMatrix = FileSystem.loadDistanceMatrix(assn.getTuple(), False)
subIdMap = FileSystem.loadSubmissionIdMap(assn.getTuple())
astNetwork = AstNetwork(assn.getTuple, distanceMatrix, subIdMap)
D = astNetwork.getDistanceList(sources, corrects)
for s in D:
nn[s] = D[s][np.argmin([y for (x, y) in D[s]])][0]
return nn
def run(self):
self.initializeLog()
for (h, p) in self.getAllParts():
assn = Assignment(h, p)
logging.info('PrecomputeNN (hw,part): ' + str(assn))
corrects = self.getASTs(assn, 'corrects')
incorrects = self.getASTs(assn, 'incorrects')
distanceMatrix = FileSystem.loadDistanceMatrix(
assn.getTuple(), False)
subIdMap = FileSystem.loadSubmissionIdMap(assn.getTuple())
astNetwork = AstNetwork(assn.getTuple(), distanceMatrix, subIdMap)
NNmap = self.getNN(corrects, incorrects, astNetwork)
outputDir = os.path.join(FileSystem.getDataDir(),
'nearestNeighbors')
if not os.path.exists(outputDir):
os.makedirs(outputDir)
outputPath = os.path.join(outputDir, 'NNmap_' + str(assn) + '.txt')
self.writeNN(outputPath, NNmap)
def getGraph(self, astSet, assn, threshold, label):
part = assn.getTuple()
#filteredGraph = nx.Graph()
filteredGraph = igraph.Graph()
distanceMatrix = FileSystem.loadDistanceMatrix(part, False)
subIdMap = FileSystem.loadSubmissionIdMap(part)
lookup = {}
for key, idx in zip(subIdMap, range(len(subIdMap))):
if int(key) in astSet:
numStudents = len(subIdMap[key])
#filteredGraph.add_node(key, {'weight': numStudents})
filteredGraph.add_vertex(label=key, weight=numStudents)
lookup[key] = filteredGraph.vs.find(label=int(key))
row = 0
toAdd = {}
while True:
logging.info('assn: ' + str(assn) + ', ' + label + ', row: ' +
str(row))
line = distanceMatrix.readline()
if not line: break
if not row in astSet:
row += 1
continue
rowValues = map(int, line.strip().split())
for col in range(row + 1, len(rowValues)):
if not col in astSet:
continue
value = rowValues[col]
if value >= 0 and value <= threshold:
toAdd[(lookup[row], lookup[col])] = value
#filteredGraph.add_edge(row, col, {'edits': value})
row += 1
logging.info('Oh... one more thing.')
filteredGraph.add_edges(toAdd.keys())
filteredGraph.es['edits'] = toAdd.values()
return filteredGraph
def getSubmissionIdMap(self, part):
return FileSystem.loadSubmissionIdMap(part)