def get_avg_degree_by_department(g, nbr_direction='in'):
'''Given a set of edges, calculate the average weight by department.
This assumes that there is some particular node of interest and that
these are neighbors of a 'target' node. See usage of function in
node_department_out_degree and node_department_in_degree for details.
params:
- edge_data (edge (tuple), weight (float)) describes edge interaction
- nbr_direction (string) target node relationship with neighbor
'''
edge_data = [((course_id, i), g.get_edge_data(course_id, i)['weight'])
for i in out_nbrs]
sum_count = defaultdict(int)
sum_weight = defaultdict(float)
for edge, weight in edge_data:
if nbr_direction == 'in':
nbr_department = get_prefix(edge[0]) # neighbor -> target
elif nbr_direction == 'out':
nbr_department = get_prefix(edge[1])
sum_count[nbr_department] += 1
sum_weight[nbr_department] += weight
avg_weight = {}
for department in sum_count.keys():
avg_weight[department] = sum_weight[department] / \
sum_count[department]
return avg_weight
def extract_departments(self):
self.department_clusters = dict()
prefixes = {get_prefix(course) for course in self.g.nodes}
for prefix in prefixes:
cluster = {
course for course in self.g.nodes
if get_prefix(course) == prefix
}
self.department_clusters[prefix] = cluster
def extract_departments(g):
department_clusters = {}
prefixes = {get_prefix(course) for course in g.nodes}
for prefix in prefixes:
cluster = {
course
for course in g.nodes if get_prefix(course) == prefix
}
department_clusters[prefix] = cluster
return department_clusters
def stopgap_nodes(g, department='', k=20):
department_node_ids = [
node for node in g.nodes if department == get_prefix(node)
]
res = []
for node_id in department_node_ids:
print(node_id)
# out degree leading into the department
out_degree_internal = sum([
i[2] for i in nx.edge_boundary(
g, [node_id], department_node_ids, data='weight')
])
sample_edges = [edge for edge in g.edges if edge[1] == node_id]
if len(sample_edges) == 0:
continue
sample_edge = sample_edges[0]
count = g.get_edge_data(sample_edge[0], sample_edge[1])['count_course']
if out_degree_internal == 0:
out_degree_internal = 1
print(
f"{int(out_degree_internal)} total {department} courses enrolled by {int(count)} students after course {node_id}"
)
coeff = out_degree_internal / count
res.append((node_id, coeff))
res = sorted(res, key=lambda x: x[1], reverse=True)[:k]
return res
def num_courses(g, k=20):
res = Counter([get_prefix(course) for course in g.nodes])
res = sorted(list(num_courses.items()), key=lambda x: x[1],
reverse=True)[:k]
return res
def plot_pagerank_by_department():
k = 9
years = ['2010-2018'] # ['2000-2008', '2005-2013', '2010-2018']
experiments = ['experiments/final/pagerank_ug_counts'
] # 'experiments/final/pagerank_ba_counts']
comparison_type = 'pagerank.json'
plt.figure(figsize=(15, 10))
the_grid = GridSpec(len(experiments), len(years))
the_grid.update(wspace=0.025, hspace=0.05)
for i, experiment_dir in enumerate(experiments):
for j, year in enumerate(years):
plt.subplot(the_grid[i, j], aspect=1)
with open(os.path.join(experiment_dir, year, comparison_type),
'r') as f:
d = json.load(f)
departments = defaultdict(float)
for entry, pagerank in d.items():
is_lang = fnmatch.filter(entry, '*LANG*') or fnmatch.filter(
entry, '*LNG*')
if is_lang:
prefix = 'LANG'
else:
prefix = get_prefix(entry)
departments[prefix] += pagerank
departments = sorted(departments.items(),
key=lambda x: x[1],
reverse=True)
top_k = departments[:k]
other = [
('OTHER',
sum([pagerank for department, pagerank in departments[k:]]))
]
groups = (top_k + other)
labels, values = zip(*groups)
print(sum(d.values()))
values = np.array(values) * 10
my_circle = plt.Circle((0, 0), 0.4, color='white')
plt.pie(values,
labels=labels,
colors=sns.light_palette(
'#99CCFF' if i == 0 else '#FC0E1C',
n_colors=k + 1,
reverse=True),
wedgeprops={
"edgecolor": "white",
'linewidth': 3,
'antialiased': True
},
textprops={'fontsize': 10})
p = plt.gcf()
p.gca().add_artist(my_circle)
if i == 0:
plt.title(year)
# plt.savefig('figures/figures_pdf/2010-2018-pr.pdf')
plt.show()
def num_courses(self):
res = Counter([get_prefix(course) for course in self.g.nodes])
res = sorted(
list(num_courses.items()), key=lambda x: x[1], reverse=True
)[:self.top_k]
return res