def distribution():
data = loader.loadAllAndClean("gpw_2007_list.txt")
networks = analyze.createNetworksSeries(data, start_date=datetime.date(year=2005, month=1, day=3), end_date=datetime.date(year=2009, month=12, day=31))
dates = pd.DatetimeIndex(networks.index)
a = 0
for i in dates:
G = nx.Graph()
if a < 40:
distance = analyze.calculateDistances(data[a:a + 40])
else:
distance = analyze.calculateDistances(data[a-20:a+20])
a +=1
for n in distance.index:
for j in distance.index:
if n != j:
if G.has_edge(n,j):
continue
else:
G.add_edge(n,j,weight = distance[n][j])
try:
network2 = mst(G)
except: ValueError
degrees = [len(list(network2.neighbors(n))) for n in network2.nodes()]
for r in degrees:
for s in z.keys():
if s == r:
z[s] +=1
if r not in z.keys():
z[r] = 1
def mol():
data = loader.loadAllAndClean("gpw_2007_list.txt")
networks = analyze.createNetworksSeries(data,
start_date=datetime.date(year=2005,
month=1,
day=3),
end_date=datetime.date(year=2009,
month=12,
day=31))
dates = pd.DatetimeIndex(networks.index)
a = 0
for i in dates:
G = nx.Graph()
apex = []
x.append(i)
if a < 30:
distance = analyze.calculateDistances(data[a:a + 30])
else:
distance = analyze.calculateDistances(data[a - 15:a + 15])
a += 1
for n in distance.index:
for j in distance.index:
if n != j:
if G.has_edge(n, j):
continue
else:
G.add_edge(n, j, weight=distance[n][j])
try:
network2 = mst(G)
for l in nx.degree(network2):
apex.append(l[0])
q = max(nx.degree_centrality(network2).values())
for key in nx.degree_centrality(network2).keys():
if nx.degree_centrality(network2)[key] == q:
center = key
way = 0
for e in apex:
way += len(nx.shortest_path(network2, center, e)) - 1
mol = way // len(apex)
except ValueError:
mol = y[-1]
y.append(mol)
def entropy():
data = loader.loadAllAndClean("gpw_2007_list.txt")
networks = analyze.createNetworksSeries(data,
start_date=datetime.date(year=2005,
month=1,
day=1),
end_date=datetime.date(year=2009,
month=12,
day=31))
dates = pd.DatetimeIndex(networks.index)
a = 0
for i in dates:
G = nx.Graph()
apex = []
x.append(i)
if a < 30:
distance = analyze.calculateDistances(data[a:a + 30])
else:
distance = analyze.calculateDistances(data[a - 15:a + 15])
for n in distance.index:
for j in distance.index:
if n != j:
if G.has_edge(n, j):
continue
else:
G.add_edge(n, j, weight=distance[n][j])
try:
network2 = mst(G)
for n in network2.nodes():
for j in network2.nodes():
if n != j:
if distance[n][j] not in apex:
if distance[j][n] not in apex:
apex.append(distance[n][j])
entrop = sts.entropy(
apex
) #próbowałem policzyć to "ręcznie" - wychodzi tak samo, ale wartości oscylują w okolicy 0,5, więc tu jest bliżej tego co na wykładzie
except ValueError:
entrop = y[-1]
y.append(entrop)
a += 1
# helper function for smooth animation - calculates nodes positions in transition frames
def averageDict(d1: dict, d2: dict, alpha):
return {n: (1 - alpha) * d1[n] + alpha * d2[n] for n in d1.keys()}
# reduces the graph
def mst(G):
return nx.algorithms.tree.mst.minimum_spanning_tree(G)
# load the data from downloaded stooq files and construct correlation networks (for now, full networks, no mst applied)
data = loader.loadCountAndClean("gpw_2007_list.txt", 30)
networks = analyze.createNetworksSeries(data,
start_date=datetime.date(year=2007,
month=12,
day=1),
end_date=datetime.date(year=2008,
month=12,
day=1))
dates = pandas.DatetimeIndex(networks.index)
# prepare subplots for plotting. ax_graph is the area where we visualize the graph (top). ax_plot is the area for plotting how measured values change (bottow)
fig = plt.figure()
gs = matplotlib.gridspec.GridSpec(2, 1, height_ratios=[3, 1])
ax_graph = fig.add_subplot(gs[0])
ax_plot = fig.add_subplot(gs[1])
# here we will store measured values in every step
values = []
# this means, transition between two steps (two networks) takes 10 frames of animation