Merge pull request #22 from ComplexData-MILA/dtdg

Dtdg

README.md

@@ -1,5 +1,14 @@
 # Temporal Graph Analysis with TGX (to appear in WSDM 2024)
 
+
+### Example Usage ###
+TGX provides many built in datasets as well as supporting TGB datasets. In addition, TGX provides dataset discretization and visualization.
+To get started, see our [starting example](https://github.com/ComplexData-MILA/TGX/blob/master/starting_example.py)
+```
+python starting_example.py -d tgbl-wiki -t daily
+```
+
+
 ### Install dependency
 Our implementation works with python >= 3.9 and can be installed as follows
 

examples/data_viz.py

@@ -46,6 +46,9 @@
 # Number of Connected Components
 tgx.connected_components_per_ts(dtdg, network_name=dataset.name)
 
+# Degree Density
+tgx.degree_density(dtdg, k=3, network_name=dataset.name)
+
 # Size of Largest Connected Component
 component_sizes = tgx.size_connected_components(dtdg)
 largest_component_sizes = [max(inner_list) if inner_list else 0 for inner_list in component_sizes]
@@ -57,5 +60,3 @@
 filename = f"{dataset.name}_average_node_engagement"
 plot_for_snapshots(engagements, y_title="Average Engagement", filename="./"+filename)
 
-# Degree Density
-tgx.degree_density(dtdg, k=3, network_name=dataset.name)

starting_example.py

@@ -1,4 +1,5 @@
 import tgx
+from tgx.utils.plotting_utils import plot_for_snapshots
 import argparse
 import sys
 
@@ -29,6 +30,50 @@ def get_args():
 # ctdg.save2csv("ctdg") #! save the graph to csv files
 
 time_scale = args.time #"minutely"  #"monthly" #"weekly" #"daily"  #"hourly" 
-dtdg = ctdg.discretize(time_scale=time_scale)
+dtdg = ctdg.discretize(time_scale=time_scale)[0]
 print ("discretize to ", time_scale)
 
+
+
+#* plotting the statistics
+tgx.degree_over_time(dtdg, network_name=dataset.name)
+tgx.nodes_over_time(dtdg, network_name=dataset.name)
+tgx.edges_over_time(dtdg, network_name=dataset.name)
+tgx.nodes_and_edges_over_time(dtdg, network_name=dataset.name)
+
+tgx.TET(dtdg, 
+        network_name=dataset.name, 
+        figsize = (9, 5),
+        axis_title_font_size = 24,
+        ticks_font_size = 24)
+
+
+tgx.TEA(dtdg, 
+        network_name=dataset.name)
+
+
+
+#* compute statistics
+test_ratio = 0.15
+tgx.get_reoccurrence(ctdg, test_ratio=test_ratio)
+tgx.get_surprise(ctdg, test_ratio=test_ratio)
+tgx.get_novelty(dtdg)
+
+
+# Number of Connected Components
+tgx.connected_components_per_ts(dtdg, network_name=dataset.name)
+
+# Degree Density
+tgx.degree_density(dtdg, k=3, network_name=dataset.name)
+
+# Size of Largest Connected Component
+component_sizes = tgx.size_connected_components(dtdg)
+largest_component_sizes = [max(inner_list) if inner_list else 0 for inner_list in component_sizes]
+filename = f"{dataset.name}_largest_connected_component_size"
+plot_for_snapshots(largest_component_sizes, y_title="Size of Largest Connected Component", filename="./"+filename)
+
+# Average Node Engagement
+engagements = tgx.get_avg_node_engagement(dtdg)
+filename = f"{dataset.name}_average_node_engagement"
+plot_for_snapshots(engagements, y_title="Average Engagement", filename="./"+filename)
+