analysis

analysis_firm_network.py

@@ -0,0 +1,101 @@
+import pandas as pd
+import matplotlib.pyplot as plt
+import networkx as nx
+import math
+
+plt.rcParams['font.sans-serif'] = 'SimHei'
+
+# count firm category
+count_firm = pd.read_csv("analysis\\count_firm.csv")
+count_firm = count_firm[count_firm['count'] > 4]
+print(count_firm.describe())
+
+count_dcp = pd.read_csv("analysis\\count_dcp.csv",
+                        dtype={
+                            'up_id_firm': str,
+                            'down_id_firm': str
+                        })
+# print(count_dcp)
+count_dcp = count_dcp[count_dcp['count'] > 2]
+
+list_firm = count_dcp['up_id_firm'].tolist(
+) + count_dcp['down_id_firm'].tolist()
+list_firm = list(set(list_firm))
+
+# init graph firm
+Firm = pd.read_csv("Firm_amended.csv")
+Firm['Code'] = Firm['Code'].astype('string')
+Firm.fillna(0, inplace=True)
+Firm_attr = Firm.loc[:, ["Code", "Name", "Type_Region", "Revenue_Log"]]
+firm_product = []
+for _, row in Firm.loc[:, '1':].iterrows():
+    firm_product.append(row[row == 1].index.to_list())
+Firm_attr.loc[:, 'Product_Code'] = firm_product
+Firm_attr.set_index('Code', inplace=True)
+
+G_firm = nx.MultiDiGraph()
+G_firm.add_nodes_from(list_firm)
+
+firm_labels_dict = {}
+for code in G_firm.nodes:
+    firm_labels_dict[code] = Firm_attr.loc[code].to_dict()
+nx.set_node_attributes(G_firm, firm_labels_dict)
+
+count_max = count_dcp['count'].max()
+count_min = count_dcp['count'].min()
+k = 5 / (count_max - count_min)
+for _, row in count_dcp.iterrows():
+    # print(row)
+    lst_add_edge = [(
+        row['up_id_firm'],
+        row['down_id_firm'],
+        {
+            'up_id_product': row['up_id_product'],
+            'up_name_product': row['up_name_product'],
+            'down_id_product': row['down_id_product'],
+            'down_name_product': row['down_name_product'],
+            # 'edge_label': f"{row['up_id_product']} {row['up_name_product']} - {row['down_id_product']} {row['down_name_product']}",
+            'edge_label': f"{row['up_id_product']} - {row['down_id_product']}",
+            'edge_width': k * (row['count'] - count_min),
+            'count': row['count']
+        })]
+    G_firm.add_edges_from(lst_add_edge)
+
+# dcp_networkx
+pos = nx.nx_agraph.graphviz_layout(G_firm, prog="dot", args="")
+node_label = nx.get_node_attributes(G_firm, 'Name')
+# node_degree = dict(G_firm.out_degree())
+node_label = {
+    # key: f"{node_label[key]} {node_degree[key]}"
+    key: f"{node_label[key]}"
+    for key in node_label.keys()
+}
+node_size = list(nx.get_node_attributes(G_firm, 'Revenue_Log').values())
+node_size = list(map(lambda x: x**2, node_size))
+edge_label = nx.get_edge_attributes(G_firm, "edge_label")
+edge_label = {(n1, n2): label for (n1, n2, _), label in edge_label.items()}
+edge_width = nx.get_edge_attributes(G_firm, "edge_width")
+edge_width = [w for (n1, n2, _), w in edge_width.items()]
+colors = nx.get_edge_attributes(G_firm, "count")
+colors = [w for (n1, n2, _), w in colors.items()]
+vmin = min(colors)
+vmax = max(colors)
+cmap = plt.cm.Blues
+fig = plt.figure(figsize=(10, 8), dpi=300)
+nx.draw(G_firm,
+        pos,
+        node_size=node_size,
+        labels=node_label,
+        font_size=6,
+        width = 3,
+        edge_color=colors,
+        edge_cmap=cmap,
+        edge_vmin=vmin,
+        edge_vmax=vmax)
+nx.draw_networkx_edge_labels(G_firm, pos, edge_label, font_size=6)
+sm = plt.cm.ScalarMappable(cmap=cmap, norm=plt.Normalize(vmin=vmin, vmax=vmax))
+sm._A = []
+position=fig.add_axes([0.9, 0.05, 0.01, 0.3])
+plt.colorbar(sm, fraction=0.01, cax=position)
+# plt.savefig("analysis\\count_dcp_network")
+plt.close()