def create_graph(messages: list, people: list, reacts: dict, freq: dict,
config: dict):
directed = config['graph']['directed']
if directed:
reacts_g = nx.DiGraph()
else:
reacts_g = nx.Graph()
reacts_g.add_nodes_from(people)
threshold = config['graph']['pair_threshold']
for pair in itertools.combinations(people, 2):
make_edge(pair[0], pair[1], reacts, directed, threshold, reacts_g)
if directed:
make_edge(pair[1], pair[0], reacts, directed, threshold, reacts_g)
vis = Network(bgcolor='#222222',
font_color="white",
height="100%",
width="100%",
directed=directed)
vis.from_nx(reacts_g)
neighs = vis.get_adj_list()
for edge in vis.edges:
edge["value"] = get_pair_reacts(edge["from"], edge["to"], reacts,
directed)
for node in vis.nodes:
neighbors = vis.neighbors(node['id'])
n_metadata = []
for neighbor in neighbors:
pair = (node['id'], neighbor)
matching_edge = [
edge for edge in vis.edges
if (edge["from"] in pair and edge["to"] in pair)
][0]
n_metadata.append(f'{neighbor} ({matching_edge["value"]})')
node['title'] = " Reacts:<br>" + "<br>".join(list(n_metadata))
if config['graph']['adjust_node_size']:
node['size'] = config['graph']['base_size'] * log10(
(freq[node['id']]) / 10)
else:
node['size'] = config['graph']['base_size']
return (vis)
def make_graph(subjects: list):
# read data
df = pd.DataFrame()
for subject in subjects:
courses = pd.read_json(f"courses/{subject}.json")
df = pd.concat([df, courses])
edge_list = make_edge_list(df)
node_list, title_list = make_node_list(df)
# make graph
g = Network(directed=True,
height="650px",
width="100%",
bgcolor="#222222",
font_color="white")
g.barnes_hut() # spring physics on the edges
g.inherit_edge_colors(False)
g.add_nodes(node_list, title=title_list)
for edge in edge_list:
g.add_edge(edge[0], edge[1], color="#94c4fc")
# add neighbor data to node hover data
for node in g.nodes:
prereq = df[df["label"] == node["label"]]["prereq"]
prereq = [] if prereq.empty else prereq.item()
next_courses = g.neighbors(node["label"])
node["title"] += "<br>Prerequisites:<br>" \
+ "<br>".join(prereq) \
+ "<br>Next courses:<br>" \
+ "<br>".join(next_courses)
node["value"] = len(next_courses) + len(prereq)
# highlight the node if it serves as a prerequisites for more than 5 course
node["font"]["size"] = node["value"] * 5
if node["value"] >= 8:
node["color"] = "red"
return g
jobtype = data['Type']
color = {'PhD' : 'blueviolet' , 'postdoc' : 'magenta' , 'link' : 'turquoise'}
all_people = set(list(people)+list(supervisors))
#node_size = Counter(supervisors)
for person in all_people:
net.add_node(person, title=person,size=5) #,shape='ellipse')
for person, supervisor, job in zip(people,supervisors, jobtype):
if job != 'link':
net.add_edge(supervisor, person, arrow=True,color=color[job])
neighbor_map = net.get_adj_list()
# add neighbor data to node hover data
for node in net.nodes:
supervised = [x for x in net.neighbors(node["id"])]
num = len(supervised)
if num > 0:
node["title"] += " Students:<br>" + "<br>".join(neighbor_map[node["id"]])
node["value"] = num
for person, supervisor, job in zip(people,supervisors, jobtype):
if job == 'link':
net.add_edge(supervisor, person, arrow=False,color=color[job],arrowStrikethrough=True)
net.add_edge(person, supervisor, arrow=False,color=color[job],arrowStrikethrough=True)
net.show("index.html")
def map_kcs(kc_list,
kc_matrix,
node_color="#8B008B",
edge_color="#03DAC6",
node_shape="ellipse",
alg="barnes",
edge_smooth=None,
buttons=False,
treshold=0.5):
"""
Use this on a test that has less than 26 KCs, or else there will be problems with
relations between the edges due to name conflict
:param threshold: float,
:param kc_list: np.array, list of all kcs
:param edge_smooth: string, How the user wants the edges to be, default is continuous
:param buttons: bool, buttons to edit graph with
:param node_shape: string, shape of node
:param edge_color: string, shape of edge
:param node_color: string, hexvalue of color for node
:param kc_matrix: 2D np array, mapping of kcs
:param alg: string, algorithm for graph
:return: renders a graph in which you can see the relations between nodes and their edges
"""
g = Network(height="1500px",
width="75%",
bgcolor="#222222",
font_color="white",
directed=True)
# if buttons:
g.width = "75%"
# nodes, layout, interaction, selection, renderer, physics
g.show_buttons(filter_=["edges", "physics"])
# Create the nodes
for kc_node in kc_list:
g.add_node(n_id=kc_node,
label=kc_node,
color=node_color,
shape=node_shape)
# Creates edges between nodes if they're connected (if the value is true in the database)
n = len(kc_matrix)
for r in range(n):
for c in range(n):
if r != c:
# Directly logically connnected edge
if 0.9 <= kc_matrix[r, c] <= 1.0:
g.add_edge(kc_list[r], kc_list[c],
color="#42cc14") # green
# Necessary edge
elif 0.7 <= kc_matrix[r, c] < 0.9:
g.add_edge(kc_list[r], kc_list[c],
color="#ff8c00") # orange
# Important edge
elif 0.5 <= kc_matrix[r, c] < 0.7:
g.add_edge(kc_list[r], kc_list[c],
color="#ffd900") # orange
elif kc_matrix[r, c] == -1.0:
g.add_edge(kc_list[r], kc_list[c], color="#56f0eb")
map_algs(g, alg)
if edge_smooth is not None:
map_smoothenes(graph=g, smooth=edge_smooth)
node_list = g.get_nodes()
for node in g.nodes:
neighbors = g.neighbors(node_list[node_list.index(node['id'])])
if neighbors is None:
title = "Endpoint KC"
else:
title = "What to learn next:<br>"
for neighbor in neighbors:
title += f"{neighbor}<br>"
node["title"] = title
# g.save_graph("../../kc.html")
g.show("kc_map.html")
