def draw_referenced_subgraph(g, website_url, thank_you_page_url, since,
min_visited_count):
raw_graph = _build_networkx_graph(g, website_url, thank_you_page_url,
since)
persistent_nodes = [
node for node, attr in raw_graph.nodes(data=True)
if attr["label"] == "persistentId"
]
graph_with_pos_computed = drawing.layout(
raw_graph,
nx.shell_layout,
nlist=[[website_url],
[
node for node, attr in raw_graph.nodes(data=True)
if attr["label"] == "transientId"
],
[
node for node, attr in raw_graph.nodes(data=True)
if attr["label"] == "persistentId"
], [thank_you_page_url]])
# update positions and change node label
raw_graph.nodes[thank_you_page_url]["pos"] += (0, 0.75)
for node in persistent_nodes:
has_visited_thank_you_page = False
visited_at_least_X_times = False
for check_name, value in raw_graph.nodes[node]["visited_events"].items(
):
if ">=" in check_name and value > 0:
if "thank" in check_name:
has_visited_thank_you_page = True
elif value > min_visited_count:
visited_at_least_X_times = True
if (has_visited_thank_you_page or not visited_at_least_X_times):
for _, to in raw_graph.edges(node):
raw_graph.nodes[to]["opacity"] = 0.25
raw_graph.nodes[node]["opacity"] = 0.25
drawing.draw(
title=
"User devices that visited ecommerce websites and optionally converted",
scatters=[drawing.edges_scatter(graph_with_pos_computed)] + list(
drawing.scatters_by_label(graph_with_pos_computed,
attrs_to_skip=["pos", "opacity"],
sizes={
"transientId": 10,
"transientId-audience": 10,
"persistentId": 20,
"persistentId-audience": 20,
"website": 30,
"thankYouPage": 30,
})))
def compute_subgraph_pos(query_results, thank_you_page):
"""Given query results compute subgraph positions."""
for persistent_id, raw_graph in _build_networkx_graph_single(
query_results=query_results,
thank_you_page=thank_you_page,
max_ts_delta=300):
raw_graph.nodes[thank_you_page]["label"] = "thank-you-page"
graph_with_pos_computed = drawing.layout(raw_graph, _custom_layout)
yield persistent_id, graph_with_pos_computed
def draw_referenced_subgraph(g, root_url):
graph = _build_networkx_graph(
root_url,
query_results=_get_transient_ids(
_get_persistent_ids_which_visited_website(g, root_url),
root_url).next())
graph = drawing.layout(graph, nx.kamada_kawai_layout)
drawing.draw(
title="Brand interaction",
scatters=[drawing.edges_scatter(graph)] + list(
drawing.scatters_by_label(graph,
attrs_to_skip=["pos"],
sizes={
"websiteGroup": 30,
"transientId": 10,
"persistentId": 15,
"website": 10
})),
)
def draw_refrenced_subgraph(g, transient_id):
raw_graph = _build_networkx_graph(
g,
g.V(transient_id).in_("has_identity").in_("member").next())
graph_with_pos_computed = drawing.layout(raw_graph,
nx.spring_layout,
iterations=2500)
drawing.draw(
title="Part of single household activity on the web",
scatters=[drawing.edges_scatter(graph_with_pos_computed)] + list(
drawing.scatters_by_label(graph_with_pos_computed,
attrs_to_skip=["pos"],
sizes={
"identityGroup": 30,
"transientId": 15,
"persistentId": 20,
"websiteGroup": 15,
"website": 10
})),
)
def draw_referenced_subgraph(g,
website_url,
categories_limit=3,
search_time_limit_in_seconds=15):
average_profile = _get_categories_popular_across_audience_of_website(
g, website_url, categories_limit=categories_limit).toList()
average_profile = dict(
chain(*category.items()) for category in average_profile)
similar_audience = _query_users_activities_stats(
g,
website_url,
average_profile,
search_time_limit_in_seconds=search_time_limit_in_seconds)
similar_audience = similar_audience.limit(15).toList()
graph = _build_graph(average_profile, similar_audience)
iabs = [
n for n, params in graph.nodes(data=True) if params["label"] == "IAB"
]
avg_iabs = [
n for n in iabs if graph.node[n]["category"] in average_profile
]
graph_with_pos_computed = drawing.layout(
graph,
nx.shell_layout,
nlist=[
["averageBuyer"],
avg_iabs,
set(iabs) - set(avg_iabs),
[
n for n, params in graph.nodes(data=True)
if params["label"] == "persistentId"
],
[
n for n, params in graph.nodes(data=True)
if params["label"] == "transientId"
],
])
# update positions
for name in set(iabs) - set(avg_iabs):
node = graph_with_pos_computed.node[name]
node["pos"] = [node["pos"][0], node["pos"][1] - 1.75]
for name in ["averageBuyer"] + avg_iabs:
node = graph_with_pos_computed.node[name]
node["pos"] = [node["pos"][0], node["pos"][1] + 1.75]
node = graph_with_pos_computed.node["averageBuyer"]
node["pos"] = [node["pos"][0], node["pos"][1] + 1]
drawing.draw(
title=
"User devices that visited ecommerce websites and optionally converted",
scatters=list(
drawing.edge_scatters_by_label(
graph_with_pos_computed,
dashes={
"interestedInButNotSufficient": "dash",
"interestedIn": "solid"
})) + list(
drawing.scatters_by_label(graph_with_pos_computed,
attrs_to_skip=["pos", "opacity"],
sizes={
"averageBuyer": 30,
"IAB": 10,
"persistentId": 20
})))