QuestionId = X.loc[:, 'QuestionId'].drop_duplicates()
QuestionId = QuestionId[QuestionId > 0]
# get data_loader
from sklearn.model_selection import KFold
kf = KFold(n_splits=5, shuffle=True)
Dataset = np.array(QuestionId[QuestionId != 0])
# get adj
edges = dataset.getAdjList_clique()
edges2 = dataset.getAdjList_Similarity1()
edges3 = dataset.getAdjList_Similarity2()
#edges4 = dataset.getAdjList_Similarity3(Pairs_train.values.tolist())
edges5 = dataset.getAdjMatrix_Identity(len(X))
Adj, rowsum = get_GCN_inputs(edges, len(X))
Adj2, rowsum = get_GCN_inputs3(edges2, len(X))
Adj3, rowsum = get_GCN_inputs3(edges3, len(X))
#Adj4, rowsum = get_GCN_inputs3(edges4, len(X))
Adj5, rowsum = get_GCN_inputs2(edges5, len(X))
Adj = Adj.to(device)
Adj2 = Adj2.to(device)
Adj3 = Adj3.to(device)
Adj5 = Adj5.to(device)
# setting of GCN
nComm = 1
nHid1 = 50
nHid2 = 30
nHid3 = 20
nHid4 = 10
#gcn_model.load_state_dict(torch.load('gcn_complete2.pkl'))
## PreProcess the dataset
cols = X.columns.drop('QTags')
X[cols] = X[cols].apply(pd.to_numeric, errors='coerce')
X.fillna(0, inplace=True)
X["PairId"] = X["PairId"].apply(lambda x: int(x)-1)
print len(X)
#print X
#edges =dataset.getAdjList_tags()
#edges = dataset.getAdjList_allTags()
#edges = dataset.getAdjList_lineGraph()
edges2 = dataset.getAdjList_clique()
edges = dataset.getAdjMatrix_Identity(len(X))
#edges = np.concatenate((edges1,edges2))
Adj, rowsum = get_GCN_inputs2(edges, len(X))
Adj2, rowsum = get_GCN_inputs(edges2, len(X))
Adj3 = makeAdj3(edges2, len(X))
#Adj2 = get_GCN_inputs(edges2, len(X))
#Adj, rowsum = normalize(Adj+Adj2)
#Adj = sparse_mx_to_torch_sparse_tensor(Adj)
#print "Adjacency Graph", Adj
#print X
data,X_Tags_Feature2 = getPostContexts(X, dataset)
#data = SparseMM(Adj)(data)
X_Tags_Feature = Variable(data, requires_grad=False)
#X_Tags_Feature = X_Tags_Feature[:,[0,3,4,5,6,7,8,9,10,13,14,15,16,17,18,19,26,27,28,29,30,31]]
#exit()