def load_physician_referral_data(infilename):
"""
Load the US physician referral data from specified zipfile
Parameters:
infilename - path name of zipflie to load from
Return value:
SNAP TNGraph object built from the data
"""
tmpdir = tempfile.mkdtemp()
try:
archive = zipfile.ZipFile(infilename, 'r')
archive.extract('physician-shared-patient-patterns-2014-days30.txt',
tmpdir)
filename = os.path.join(
tmpdir, "physician-shared-patient-patterns-2014-days30.txt")
archive.close()
context = snap.TTableContext()
schema = snap.Schema()
## schema.Add(snap.TStrTAttrPr("NPI_1", snap.atInt))
## schema.Add(snap.TStrTAttrPr("NPI_2", snap.atInt))
# the above 2 lines worked with SNAP 4.0.0 on VLSCI
# but now using SNAP 4.1.0
# on hpc.ics.usi.ch find that all ids are -1 so graph wrong.
# Cannot work out why so changed to string not int to try to fix it:
schema.Add(snap.TStrTAttrPr("NPI_1", snap.atStr))
schema.Add(snap.TStrTAttrPr("NPI_2", snap.atStr))
## schema.Add(snap.TStrTAttrPr("count", snap.atInt))
## schema.Add(snap.TStrTAttrPr("unique_bene", snap.atInt))
## schema.Add(snap.TStrTAttrPr("same_day_count", snap.atInt))
# The above 3 lines also worked fine with SNAP 4.0.0 before but
# now fail on SNAP 4.1.0 (seems to be due to spaces in CSV fields,
# not inexplicable like first two which have no spaces) but not using
# them at the moment anyway so easier to just make (unused) strings:
schema.Add(snap.TStrTAttrPr("count", snap.atStr))
schema.Add(snap.TStrTAttrPr("unique_bene", snap.atStr))
schema.Add(snap.TStrTAttrPr("same_day_count", snap.atStr))
table = snap.TTable.LoadSS(schema, filename, context, ",",
snap.TBool(False))
G = snap.ToGraph(snap.PNGraph, table, "NPI_1", "NPI_2", snap.aaFirst)
finally:
cleanup_tmpdir(tmpdir)
return G
import snap
graphfilename = "C:\Python27\HW1\wiki-vote.txt"
schema = snap.Schema()
context = snap.TTableContext()
schema.Add(snap.TStrTAttrPr("srcID", snap.atStr))
schema.Add(snap.TStrTAttrPr("dstID", snap.atStr))
sample_table = snap.TTable.LoadSS(schema, graphfilename, context, "\t",
snap.TBool(False))
# graph will be an object of type snap.PNGraph
graph = snap.ToGraph(snap.PNGraph, sample_table, "srcID", "dstID",
snap.aaFirst)
#no of nodes
Count = snap.CntNonZNodes(graph)
print "Count of nodes with degree greater than 0 is %d" % Count
#no of edges
Count = snap.CntOutDegNodes(graph, 0)
print "Count of nodes with out-degree 0 is %d" % Count
#no of nodes with zero in-degree
Count = snap.CntInDegNodes(graph, 0)
print "Count of nodes with in-degree 0 is %d" % Count
#no of directed edges
Count = snap.CntUniqDirEdges(graph)
print "Count of directed edges is %d" % Count
#no of undirected edges
Count = snap.CntUniqUndirEdges(graph)
print "Count of undirected edges is %d" % Count
#no of self edges
Count = snap.CntSelfEdges(graph)
print "Count of self edges is %d" % Count
# Rename
# >>> posts.rename('UserId','Expert')
posts.Rename("UserId", "Expert")
t.show("rename", posts)
# Join
# >>> edges = questions.join(posts, ['AcceptedAnswerId'], ['PostId'])
edges = questions.Join("t1.AcceptedAnswerId", posts, "PostId")
t.show("join", edges)
# Create haskell-specific Q&A graph
# >>> graph = posts.graph('Asker', 'Expert', directed = True)
edges.SetSrcCol("t1_t2.Asker")
edges.SetDstCol("t1.Expert")
graph = snap.ToGraph(edges, snap.aaFirst)
t.show("graph", graph)
# Compute Authority score
# >>> hits = graph.hits('Authority', 'Hub')
# note: the code below creates a table (Node name, Authority score) - the hub score is not used
HTHub = snap.TIntFltH()
HTAuth = snap.TIntFltH()
snap.GetHits(graph, HTHub, HTAuth)
authority = snap.TTable.New("authority", HTAuth, "Expert", AUTHORITY_ATTRIBUTE,
context, snap.TBool(False))
t.show("authority score", authority)
# b) Compute comment scores
# Load comments
# Join
# >>> t3 = t1.join(t2)
#t3 = t1.Join("PostId", t2, "PostId")
#t.show("join", t3)
# Join
# >>> t4 = t3.join(t1, ["AnswerId"], ["PostId"])
t4 = t1.Join("AnswerId", t1, "PostId")
t.show("join", t4)
# Graph
# >>> graph = t4.graph("UserId_1", "UserId_2")
t4.SetSrcCol("t1_1.UserId")
t4.SetDstCol("t1_2.UserId")
graph = snap.ToGraph(
t4, snap.aaFirst
) # ToGraphPerGroup should be able to support grouping on string columns!
t.show("graph", graph)
# Get authority scores
HTHub = snap.TIntFltH()
HTAuth = snap.TIntFltH()
snap.GetHits(graph, HTHub, HTAuth)
t.show("hits", graph)
t5 = snap.TTable.TableFromHashMap("t5", HTAuth, "UserId", "Authority", context,
snap.TBool(False))
t.show("authority score", t5)
# Select top entries
# >>> t.select('Authority > 0.0')
# Self-join
# >>> table.selfjoin(table, ['Key'])
table = table.SelfJoin("Key")
t.show("join", table)
# Select
# >>> table.select('Author_1 != Author_2')
table.SelectAtomic("1_2_1.1.Author", "1_2_2.1.Author", snap.NEQ)
t.show("select", table)
# Create network
# >>> table.graph('Author_1', 'Author_2', directed=False)
table.SetSrcCol("1_2_1.1.Author")
table.SetDstCol("1_2_2.1.Author")
graph = snap.ToGraph(table, snap.aaFirst)
t.show("graph", graph)
# Compute PageRank score
# >>> pagerank = graph.pageRank('PageRank')
HT = snap.TIntFltH()
snap.GetPageRank(graph, HT)
pagerank = snap.TTable.New("PR", HT, "Author", PAGE_RANK_ATTRIBUTE, context, snap.TBool(True))
t.show("page rank", pagerank)
# Order by PageRank score (in descending order)
# >>> pagerank.order(['PageRank'], desc = True)
V = snap.TStrV()
V.Add(PAGE_RANK_ATTRIBUTE)
pagerank.Order(V, "", snap.TBool(False), snap.TBool(False))
t.show("order", pagerank)
r.show("__references__")
# load context
print time.ctime(), "loading context ..."
context.Load(FIn)
t.show("loadbin context", RefsT)
r.show("__context__")
print time.ctime(), "done"
# In[2]:
# Create the network
refs_schema = map(lambda x: x.GetVal1(), RefsT.GetSchema())
print time.ctime(), "Creating network ..."
net = snap.ToGraph(snap.PNGraph, RefsT, refs_schema[0], refs_schema[1],
snap.aaFirst)
print time.ctime(), "done."
# In[3]:
# Compute InDegV from references graph to get number of citations of each paper.
print time.ctime(), "Computing indegv ..."
InDegV = snap.TIntPrV()
snap.GetNodeInDegV(net, InDegV)
t.show("indegv", InDegV)
r.show("__InDegV__")
print time.ctime()
# In[4]:
# Compute PageRank from references graph.
t.show("select tag = 'python'", questions)
r.show("__selecttagpython__")
questions.SelectAtomicIntConst("AcceptedAnswerId", 0, snap.NEQ)
t.show("select questions", questions)
r.show("__selectquestions__")
table.SelectAtomicIntConst("AcceptedAnswerId", 0, snap.EQ)
t.show("select answers", table)
r.show("__selectanswers__")
qa = questions.Join("AcceptedAnswerId", table, "Id")
t.show("join", qa)
r.show("__join__")
graph = snap.ToGraph(snap.PUNGraph, qa, "2.OwnerUserId", "1.OwnerUserId",
snap.aaFirst)
t.show("graph", graph)
r.show("__graph__")
PRankH = snap.TIntFltH()
snap.GetPageRank(graph, PRankH, 0.85, 1e-4, 100)
prtable = snap.TTable.New("PR", PRankH, "UserId", "PageRank", context,
snap.TBool(True))
t.show("pagerank", prtable)
r.show("__pagerank__")
FOut = snap.TFOut(dstfile)
prtable.Save(FOut)
t.show("save bin", prtable)
r.show("__savebin__")
print("G3 nodes", G3.GetNodes())
print("G3 edges", G3.GetEdges())
t = printtime(t, "saving the graph to binary")
FOut = snap.TFOut(binname)
G1.Save(FOut)
FOut.Flush()
t = printtime(t, "reading the graph from binary")
FIn = snap.TFIn(binname)
G4 = snap.TUNGraph.Load(FIn)
print("G4 nodes", G4.GetNodes())
print("G4 edges", G4.GetEdges())
t = printtime(t, "reading the graph as table")
context = snap.TTableContext()
schema = snap.Schema()
schema.Add(snap.TStrTAttrPr("SrcID", snap.atInt))
schema.Add(snap.TStrTAttrPr("DstID", snap.atInt))
T1 = snap.TTable.LoadSS(schema, txtname, context, "\t", snap.TBool(False))
print("T1 rows", T1.GetNumRows())
t = printtime(t, "converting table to graph")
G5 = snap.ToGraph(snap.PUNGraph, T1, "SrcID", "DstID", snap.aaFirst)
print("G5 nodes", G5.GetNodes())
print("G5 edges", G5.GetEdges())
printtime(t, "done")
t.show("select tag = 'python'", questions)
r.show("__selecttagpython__")
questions.SelectAtomicIntConst("AcceptedAnswerId", 0, snap.NEQ)
t.show("select questions", questions)
r.show("__selectquestions__")
table.SelectAtomicIntConst("AcceptedAnswerId", 0, snap.EQ)
t.show("select answers", table)
r.show("__selectanswers__")
qa = questions.Join("AcceptedAnswerId", table, "Id")
t.show("join", qa)
r.show("__join__")
graph = snap.ToGraph(snap.PUNGraph, qa, "OwnerUserId-2", "OwnerUserId-1",
snap.aaFirst)
t.show("graph", graph)
r.show("__graph__")
PRankH = snap.TIntFltH()
snap.GetPageRank(graph, PRankH, 0.85, 1e-4, 100)
prtable = snap.TTable.New(PRankH, "UserId", "PageRank", context,
snap.TBool(True))
t.show("pagerank", prtable)
r.show("__pagerank__")
FOut = snap.TFOut(dstfile)
prtable.Save(FOut)
t.show("save bin", prtable)
r.show("__savebin__")
# Join
# >>> t3 = t1.join(t2)
t3 = t1.Join("PostId", t2, "PostId")
t.show("join", t3)
# Join
# >>> t4 = t3.join(t1, ["AnswerId"], ["PostId"])
t4 = t3.Join("t1.AnswerId", t1, "PostId")
t.show("join", t4)
# Graph
# >>> graph = t4.graph("UserId_1", "UserId_2")
#t4.SetSrcCol("t1_t2.t1.UserId")
#t4.SetDstCol("t1.UserId")
graph = snap.ToGraph(
snap.PNGraph, t4, "t1_t2.t1.UserId", "t1.UserId", snap.aaFirst
) # ToGraphPerGroup should be able to support grouping on string columns!
t.show("graph", graph)
#graph.Dump()
# Get authority scores
HTHub = snap.TIntFltH()
HTAuth = snap.TIntFltH()
snap.GetHits(graph, HTHub, HTAuth)
t.show("hits", graph)
t5 = snap.TTable.TableFromHashMap("t5", HTAuth, "UserId", "Authority", context,
snap.TBool(False))
t.show("authority score", t5)
# Select top entries