From 36b8ca0a54a26e8e2a935f4a12948c17086552e8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E6=9D=8E=E8=A1=8C=E4=BA=91?=
 <57734284+EricLinuxLXY@users.noreply.github.com>
Date: Thu, 21 Nov 2019 14:03:18 +0800
Subject: [PATCH 1/2] Create networksynergy.py
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

对外直接投资网络协同效应：计算由于企业OFDI而导致的企业面对的贸易网络重组（每个企业由于其当前OFDI与历史OFDI而产生异质性）后结构洞（Constraints）与中心度（Eigenvector Centrality）的改变。
---
 networksynergy.py | 127 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 127 insertions(+)
 create mode 100644 networksynergy.py

diff --git a/networksynergy.py b/networksynergy.py
new file mode 100644
index 0000000..383e3f6
--- /dev/null
+++ b/networksynergy.py
@@ -0,0 +1,127 @@
+from pandas import DataFrame, Series
+from collections import defaultdict
+import community
+import pandas as pd; import numpy as np
+import csv #使用内置的CSV库
+import os
+import matplotlib.pyplot as plt
+import networkx as nx
+
+os.chdir("/Users/eric/Documents/statwork/NetworkSynergy/NetSynergyChange")
+cwd = os.getcwd()
+
+"""第二阶段：将配对后的重构信息转换格式导入网络图"""
+"""将包含历史信息的数据处理，将档期OFDI与历史OFDI配对"""
+treateddf = pd.read_stata('NetsynergyInformation.dta', index_col = 'groupkeys')
+
+temporatyofdis = treateddf[['siren', 'year', 'paysnew', 'paysold']]
+
+temporatyofdis = temporatyofdis.drop_duplicates()
+temporatyofdis.head()
+
+temporatyofdis[-10:]
+
+"""采用遍历各分组的方式可以简单地将数据框按照分组变量拆分"""
+samples = [group for (_, _, _), group in df.groupby(['siren', 'year', 'paysnew'])]
+
+"""
+"""这里曾出现DUG：当只有一条信息时，重新索引会将一个元素变成Series，因此会产生不同的格式"""
+"""构造函数：拆分数据"""
+
+def split_by_index(index):
+    
+    temps = temporatyofdis.loc[index]
+    if type(temps) == pd.core.frame.DataFrame:
+        return temps
+    else:
+        return DataFrame(temps).T
+
+    
+"""使用函数将其拆分"""
+indexlst = list(np.unique(temporatyofdis.index))
+samples = [split_by_index(index) for index in indexlst]
+"""
+
+"""将每一个DataFrame转换成列表与元组"""
+def transforinfomation(df):
+        
+    lst = df.iloc[0, :3].values
+    results=list(lst)
+    results.append(list(zip(df['paysold'], df['paysnew'])))
+       
+    return results
+
+
+networkchangesinfo = [transforinfomation(df) for df in samples]
+
+
+locations = pd.read_stata('all_lat_lang.dta')
+
+'''输出各节点结构洞与中心度指数-list'''
+'''注意的是：由于OFDI网络协同效应（Network Synergy）重构了贸易网络，因此采用有向无权网络'''
+def NetworkChanges(changes):
+    df = pd.read_stata("./tradefiles/wtf_bilat_exptop3_%s.dta" % changes[1])
+    '''生成网络图'''    
+    '''获取边'''
+    f = df.iloc[:, 2]
+    to = df.iloc[:, 3]
+    
+    edgesList = list(zip(f, to))
+    
+    nodelist = list(locations.iloc[:,6])
+    
+    ''''''
+    '''构建有向图'''
+    G = nx.DiGraph()
+    G.add_nodes_from(nodelist)
+    """添加有向有权网络图的边"""
+    G.add_edges_from(edgesList)
+    """加入Network Synergy重构"""
+    G.add_edges_from(changes[3])
+
+    """获取网络指标并保存到数组Array中"""
+    data = np.array([changes[0]], dtype = object)
+    data = np.vstack((data, [changes[1]]))
+    data = np.vstack((data, [changes[2]]))
+    """获取网络指标"""
+    constraintdict = nx.constraint(G)
+    data = np.vstack((data, [constraintdict[changes[2]]]))
+    try:
+        """"""
+        eigenvector_centrality = nx.eigenvector_centrality(G)
+        data = np.vstack((data, [eigenvector_centrality[changes[2]]]))
+        
+    except nx.exception.PowerIterationFailedConvergence as e:
+        """如果报错，就将其特征向量中心度用缺失值填充"""
+        data = np.vstack((data, np.nan))
+
+
+    return data
+    
+"""构建一个更改数据格式的函数，以方便能够成功导出"""
+def DtaFrameshr(dta):
+    dta.columns = ['siren', 'year', 'paysnum', 'Structural_Holes_Post',
+                          'Eigenvector_Centrality_Post']
+    """将Object对象转换格式"""
+    dta['Structural_Holes_Post'] = dta['Structural_Holes_Post'].astype(float)
+    dta['Eigenvector_Centrality_Post'] = dta['Eigenvector_Centrality_Post'].astype(float)
+    return dta
+    
+import datetime
+lst = [['siren'], ['year'],['paysnum'],[100],[200]]
+dfs = np.array(lst, dtype = object)
+"""循环"""
+print(datetime.datetime.now())
+
+counts = 0
+for row in networkchangesinfo:
+    dfs = np.hstack((dfs,NetworkChanges(row)))
+    counts += 1
+    if (counts%5000) == 0 :
+        print(datetime.datetime.now())
+        print("现在循环到：%d/46171......" % counts)
+            
+synergychanges = DataFrame(dfs)
+synergychanges = synergychanges.T
+dtaframes = DtaFrameshr(synergychanges)
+dtaframes.to_stata("NSCdf.dta", write_index = False)

From fe840105eec65cff4701c4ba05dfb7e00e67e2cc Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E6=9D=8E=E8=A1=8C=E4=BA=91?=
 <57734284+EricLinuxLXY@users.noreply.github.com>
Date: Thu, 21 Nov 2019 14:28:25 +0800
Subject: [PATCH 2/2] Create .gitignore

---
 .gitignore | 1 +
 1 file changed, 1 insertion(+)
 create mode 100644 .gitignore

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..e43b0f9
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1 @@
+.DS_Store