[FRAUD] 그래프 그림 그리기(graft)

imports

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt 
import networkx as nx
import sklearn

# sklearn
from sklearn import model_selection # split함수이용
from sklearn import ensemble # RF,GBM
from sklearn import metrics
from sklearn.metrics import precision_score, recall_score, f1_score
from sklearn.svm import SVC
from sklearn.ensemble import RandomForestClassifier
from sklearn.naive_bayes import GaussianNB

# gnn
import torch
import torch.nn.functional as F
import torch_geometric
from torch_geometric.nn import GCNConv

import graft

def down_sample_textbook(df):
    df_majority = df[df.is_fraud==0].copy()
    df_minority = df[df.is_fraud==1].copy()
    df_maj_dowsampled = sklearn.utils.resample(df_majority, n_samples=len(df_minority), replace=False, random_state=42)
    df_downsampled = pd.concat([df_minority, df_maj_dowsampled])
    return df_downsampled

def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

def mask(df):
    df_tr,df_test = sklearn.model_selection.train_test_split(df, random_state=42)
    N = len(df)
    train_mask = [i in df_tr.index for i in range(N)]
    test_mask = [i in df_test.index for i in range(N)]
    train_mask = np.array(train_mask)
    test_mask = np.array(test_mask)
    return train_mask, test_mask

def edge_index_selected(edge_index):
    theta = edge_index[:,2].mean()
    edge_index[:,2] = (np.exp(-edge_index[:,2]/theta) != 1)*(np.exp(-edge_index[:,2]/theta))
    edge_index = edge_index.tolist()
    mean_ = np.array(edge_index)[:,2].mean()
    selected_edges = [(int(row[0]), int(row[1])) for row in edge_index if row[2] > mean_]
    edge_index_selected = torch.tensor(selected_edges, dtype=torch.long).t()
    return edge_index_selected

fraudTrain = pd.read_csv("~/Desktop/fraudTrain.csv").iloc[:,1:]

len(set(fraudTrain['cc_num']))

943

len(set(fraudTrain['merchant']))

693

# # fraudTrain = fraudTrain.assign(trans_date_trans_time= list(map(lambda x: pd.to_datetime(x), fraudTrain.trans_date_trans_time)))
# fraudTrain = fraudTrain.assign(
#     trans_date_trans_time= fraudTrain.trans_date_trans_time.apply(pd.to_datetime)
# )
fraudTrain = pd.read_pickle('temp.pkl')

데이터정리

_df1 = fraudTrain[fraudTrain["is_fraud"] == 0].sample(frac=0.20, random_state=42)
_df2 = fraudTrain[fraudTrain["is_fraud"] == 1]
df02 = pd.concat([_df1,_df2])
df02.shape

(214520, 22)

df50 = down_sample_textbook(df02)
df50 = df50.reset_index()
df50.shape

(12012, 23)

---

tr/test

mask(df50)

train_mask, test_mask = mask(df50)

---

---

분석 1(GCN): amt

# x = torch.tensor(df50['amt'], dtype=torch.float).reshape(-1,1)
# y = torch.tensor(df50['is_fraud'],dtype=torch.int64)
# data = torch_geometric.data.Data(x=x, edge_index = edge_index_selected, y=y, train_mask = train_mask, test_mask = test_mask)
# data



# torch.manual_seed(202250926)
# class GCN2(torch.nn.Module):
#     def __init__(self):
#         super().__init__()
#         self.conv1 = GCNConv(1, 32)
#         self.conv2 = GCNConv(32,2)

#     def forward(self, data):
#         x, edge_index = data.x, data.edge_index

#         x = self.conv1(x, edge_index)
#         x = F.relu(x)
#         x = F.dropout(x, training=self.training)
#         x = self.conv2(x, edge_index)

#         return F.log_softmax(x, dim=1)

# X = (data.x[data.train_mask]).numpy()
# XX = (data.x[data.test_mask]).numpy()
# y = (data.y[data.train_mask]).numpy()
# yy = (data.y[data.test_mask]).numpy()

# model = GCN2()
# optimizer = torch.optim.Adam(model.parameters(), lr=0.01, weight_decay=5e-4)
# model.train()
# for epoch in range(400):
#     optimizer.zero_grad()
#     out = model(data)
#     loss = F.nll_loss(out[data.train_mask], data.y[data.train_mask])
#     loss.backward()
#     optimizer.step()
# model.eval()

# pred = model(data).argmax(dim=1)
# yyhat = pred[data.test_mask]

# metrics = [sklearn.metrics.accuracy_score,
#            sklearn.metrics.precision_score,
#            sklearn.metrics.recall_score,
#            sklearn.metrics.f1_score]

# _results1= pd.DataFrame({m.__name__:[m(yy,yyhat).round(6)] for m in metrics},index=['분석1'])
# _results1

# x = torch.tensor(df50['amt'], dtype=torch.float).reshape(-1,1)
# y = torch.tensor(df50['is_fraud'],dtype=torch.int64)
# data = torch_geometric.data.Data(x=x, edge_index = edge_index_selected, y=y, train_mask = train_mask, test_mask = test_mask)
# data

unweighted graph

cc_num_counts = df50['cc_num'].value_counts()
cc_num_10_nodes = cc_num_counts[cc_num_counts == 10].index.tolist()

cc10 = df50[(df50['cc_num'].isin(cc_num_10_nodes))]

cc10.groupby("cc_num")['is_fraud'].sum().sort_values()

cc_num
4.958590e+18     0
4.855490e+18     0
3.573030e+15     0
2.254800e+15     0
4.170690e+15     0
4.708990e+15     0
3.750830e+14     0
6.011400e+15     0
2.296010e+15     0
3.023880e+13     0
3.524570e+15     0
4.783230e+12     0
4.623560e+12     0
6.596740e+15     0
6.763730e+11     0
4.761400e+18     0
3.019960e+13     0
4.809700e+12     2
4.503100e+18     2
6.511350e+15     4
5.020500e+11     5
6.593250e+15     6
6.523180e+15     7
4.996260e+15     7
6.528910e+15     7
3.554850e+15     7
2.131140e+14     7
2.274910e+15     7
3.636050e+13     7
6.304130e+11     8
4.756040e+18     8
4.605900e+12     8
6.502200e+15     8
5.018280e+11     8
4.228410e+15     8
3.772650e+14     8
4.670610e+15     8
4.358140e+18     9
4.209700e+18     9
4.824020e+15     9
6.011200e+15     9
3.462080e+14     9
3.529790e+15    10
1.800380e+14    10
4.669560e+15    10
4.844240e+12    10
3.756230e+14    10
4.110270e+18    10
4.162870e+18    10
4.257110e+18    10
5.144040e+15    10
4.457490e+12    10
2.295960e+15    10
3.452260e+14    10
3.542830e+15    10
Name: is_fraud, dtype: int64

c_sample = 4.809700e+12

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)

edge_index = (np.array(edge_index)).astype(np.float64)

theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta)!= 1)*(np.exp(-edge_index[:,2]/theta))
    

edge_index = edge_index.tolist()
mean_ = np.array(edge_index)[:,2].mean()
selected_edges = [(int(row[0]), int(row[1])) for row in edge_index if row[2] > mean_]
edge_index_selected = torch.tensor(selected_edges, dtype=torch.long).t()

edge_index_selected.shape

torch.Size([2, 42])

selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()

# 두 번째 데이터 선택
# selected_indices2 = df50[df50['cc_num'] == 4.2929e+18].index
# is_fraud_values2 = df50.loc[selected_indices2, 'is_fraud'].tolist()


x = df50.loc[selected_indices1, 'amt'].tolist()

y = df50.loc[selected_indices1, 'is_fraud'].tolist()

y

[1, 1, 0, 0, 0, 0, 0, 0, 0, 0]

g = torch_geometric.data.Data(
    edge_index = edge_index_selected,
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

g = torch_geometric.data.Data(
    edge_index = edge_index_selected,
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = edge_index_selected
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

edge_index_selected

tensor([[ 4450,  4450,  4450,  4450,  4462,  4462,  4462,  4462,  7320,  7320,
          7936,  7936,  7936,  7936,  7936,  7936,  8128,  8128,  8128,  8128,
          9654,  9654,  9654,  9654,  9668,  9668,  9668,  9668,  9668, 10079,
         10079, 10079, 10079, 11078, 11078, 11078, 11078, 11078, 11177, 11177,
         11177, 11177],
        [ 4462,  7320,  9668, 10079,  4450,  7320,  9668, 10079,  4450,  4462,
          8128,  9654,  9668, 10079, 11078, 11177,  7936,  9654, 11078, 11177,
          7936,  8128, 11078, 11177,  4450,  4462,  7936, 10079, 11078,  4450,
          4462,  7936,  9668,  7936,  8128,  9654,  9668, 11177,  7936,  8128,
          9654, 11078]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

g.y

[1, 1, 0, 0, 0, 0, 0, 0, 0, 0]

edge_index_selected2

tensor([[2, 2, 2, 2, 6, 6, 6, 6, 8, 8, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 7, 7, 7, 7,
         3, 3, 3, 3, 3, 9, 9, 9, 9, 4, 4, 4, 4, 4, 5, 5, 5, 5],
        [6, 8, 3, 9, 2, 8, 3, 9, 2, 6, 1, 7, 3, 9, 4, 5, 0, 7, 4, 5, 0, 1, 4, 5,
         2, 6, 0, 9, 4, 2, 6, 0, 3, 0, 1, 7, 3, 5, 0, 1, 7, 4]])

g.y

[1, 1, 0, 0, 0, 0, 0, 0, 0, 0]

g.x

[8.32, 17.08, 31.73, 65.4, 64.53, 70.88, 66.18, 15.09, 127.96, 20.28]

dr_opts = { 
    'output_size': (500,500)
}
graft.graph.plot_undirected_unweighted(
    g,
    node_color=g.y,
    node_size=g.x,
    draw_options= dr_opts
)

/home/coco/anaconda3/envs/test/lib/python3.10/site-packages/torch_geometric/data/storage.py:327: UserWarning: Unable to accurately infer 'num_nodes' from the attribute set '{'y', 'edge_index', 'x'}'. Please explicitly set 'num_nodes' as an attribute of 'data' to suppress this warning
  warnings.warn(

---

c_sample = 4.503100e+18

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)

edge_index = (np.array(edge_index)).astype(np.float64)

theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta)!= 1)*(np.exp(-edge_index[:,2]/theta))
    

edge_index = edge_index.tolist()
mean_ = np.array(edge_index)[:,2].mean()
selected_edges = [(int(row[0]), int(row[1])) for row in edge_index if row[2] > mean_]
edge_index_selected = torch.tensor(selected_edges, dtype=torch.long).t()

edge_index_selected.shape

selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()

# 두 번째 데이터 선택
# selected_indices2 = df50[df50['cc_num'] == 4.2929e+18].index
# is_fraud_values2 = df50.loc[selected_indices2, 'is_fraud'].tolist()


x = df50.loc[selected_indices1, 'amt'].tolist()

y = df50.loc[selected_indices1, 'is_fraud'].tolist()

g = torch_geometric.data.Data(
    edge_index = edge_index_selected,
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)


unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = edge_index_selected
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (500,500)
}
graft.graph.plot_undirected_unweighted(
    g,
    node_color=g.y,
    node_size=g.x,
    draw_options= dr_opts
)

---

c_sample = 6.511350e+15

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)

edge_index = (np.array(edge_index)).astype(np.float64)

theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta)!= 1)*(np.exp(-edge_index[:,2]/theta))
    

edge_index = edge_index.tolist()
mean_ = np.array(edge_index)[:,2].mean()
selected_edges = [(int(row[0]), int(row[1])) for row in edge_index if row[2] > mean_]
edge_index_selected = torch.tensor(selected_edges, dtype=torch.long).t()

edge_index_selected.shape

selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()

# 두 번째 데이터 선택
# selected_indices2 = df50[df50['cc_num'] == 4.2929e+18].index
# is_fraud_values2 = df50.loc[selected_indices2, 'is_fraud'].tolist()


x = df50.loc[selected_indices1, 'amt'].tolist()

y = df50.loc[selected_indices1, 'is_fraud'].tolist()

g = torch_geometric.data.Data(
    edge_index = edge_index_selected,
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)


unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = edge_index_selected
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (500,500)
}
graft.graph.plot_undirected_unweighted(
    g,
    node_color=g.y,
    node_size=g.x,
    draw_options= dr_opts
)

??

---

c_sample = 5.020500e+11

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)

edge_index = (np.array(edge_index)).astype(np.float64)

theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta)!= 1)*(np.exp(-edge_index[:,2]/theta))
    

edge_index = edge_index.tolist()
mean_ = np.array(edge_index)[:,2].mean()
selected_edges = [(int(row[0]), int(row[1])) for row in edge_index if row[2] > mean_]
edge_index_selected = torch.tensor(selected_edges, dtype=torch.long).t()

edge_index_selected.shape

selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()

# 두 번째 데이터 선택
# selected_indices2 = df50[df50['cc_num'] == 4.2929e+18].index
# is_fraud_values2 = df50.loc[selected_indices2, 'is_fraud'].tolist()


x = df50.loc[selected_indices1, 'amt'].tolist()

y = df50.loc[selected_indices1, 'is_fraud'].tolist()

g = torch_geometric.data.Data(
    edge_index = edge_index_selected,
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)


unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = edge_index_selected
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (500,500)
}
graft.graph.plot_undirected_unweighted(
    g,
    node_color=g.y,
    node_size=g.x,
    draw_options= dr_opts
)

df50[df50.cc_num.apply(int) ==  4.503100e+18]

	index	trans_date_trans_time	cc_num	merchant	category	amt	first	last	gender	street	...	lat	long	city_pop	job	dob	trans_num	unix_time	merch_lat	merch_long	is_fraud
5381	949044	2020-01-15 02:04:00	4.503100e+18	fraud_Schumm, Bauch and Ondricka	grocery_pos	332.10	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	866751490a47e75897fb852fcf3684a5	1358215494	28.306661	-81.336753	1
5382	949085	2020-01-15 03:13:00	4.503100e+18	fraud_Cremin, Hamill and Reichel	misc_pos	8.96	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	ad9d834d20eaa6c385e2da4ca8122f08	1358219588	28.258438	-81.382080	1
7996	876537	2019-12-21 17:39:00	4.503100e+18	fraud_Kihn, Brakus and Goyette	personal_care	19.74	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	806463bc8f21d4ede724ee50f5a209bc	1356111580	27.037246	-81.384770	0
8217	815021	2019-12-07 23:22:00	4.503100e+18	fraud_Pouros-Haag	shopping_pos	5.76	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	1e28476a15ad9ed769d0d71d6cc3eb02	1354922529	28.636698	-82.419218	0
9176	921518	2019-12-30 22:32:00	4.503100e+18	fraud_Schroeder Group	health_fitness	38.49	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	83cfa21de5d1366a0723351a291f761e	1356906720	27.997746	-82.168664	0
10248	418508	2019-07-07 18:01:00	4.503100e+18	fraud_Boyer PLC	shopping_net	3.49	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	453f3e6eece7279bea060dc58b765d3f	1341684100	28.687747	-82.120574	0
10517	636869	2019-09-28 13:23:00	4.503100e+18	fraud_Ruecker, Beer and Collier	shopping_net	67.81	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	46925141500a5774537115173f74e084	1348838599	27.360343	-83.002864	0
10613	138138	2019-03-16 21:05:00	4.503100e+18	fraud_Turner and Sons	shopping_pos	291.51	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	680af7c15f20b6b0e3a4acf28c8ee306	1331931943	28.430308	-81.703027	0
10950	255886	2019-05-07 07:46:00	4.503100e+18	fraud_Stracke-Lemke	grocery_pos	213.37	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	297be021b9b297f7857a29053a41c903	1336376769	27.211371	-81.875982	0
11321	611610	2019-09-16 16:54:00	4.503100e+18	fraud_Heathcote LLC	shopping_net	9.51	Katherine	Tucker	F	670 Le Meadows Suite 250	...	27.9865	-82.0139	237282	Clothing/textile technologist	1979-07-03	6705120619ad8b218cdb5dad91628cc5	1347814458	28.862677	-82.120978	0

10 rows × 23 columns

---

c_sample = 4.503100e+18

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)

edge_index = (np.array(edge_index)).astype(np.float64)

theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta)!= 1)*(np.exp(-edge_index[:,2]/theta))
    

edge_index = edge_index.tolist()

np.array(edge_index)[:,:2]  # edge!!
np.array(edge_index)[:,-1]  # weight



selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()



x = df50.loc[selected_indices1, 'amt'].tolist()

y = df50.loc[selected_indices1, 'is_fraud'].tolist()

g = torch_geometric.data.Data(
    edge_index = np.array(edge_index)[:,:2],
    edge_attr = np.array(edge_index)[:,-1],  # weight
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)




unique_nodes = set(g.edge_index.reshape(-1).tolist())

dict_map = {n:i for i,n in enumerate(unique_nodes)}

source, target = np.array(edge_index)[:,0], np.array(edge_index)[:,1]


edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    edge_attr = np.array(edge_index)[:,-1],
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (200,200),
    'edge_marker_size': 1,
  'output':f"ten-nodes-{c_sample}.pdf",
}
graft.graph.plot_undirected_weighted(
    g,
    node_color=g.y,
    node_size=g.x,
    edge_weight_text=False,
    draw_options= dr_opts,
)

---

c_sample = 6.511350e+15

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)

edge_index = (np.array(edge_index)).astype(np.float64)

theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta)!= 1)*(np.exp(-edge_index[:,2]/theta))
    

edge_index = edge_index.tolist()

np.array(edge_index)[:,:2]  # edge!!
np.array(edge_index)[:,-1]  # weight



selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()



x = df50.loc[selected_indices1, 'amt'].tolist()

y = df50.loc[selected_indices1, 'is_fraud'].tolist()

g = torch_geometric.data.Data(
    edge_index = np.array(edge_index)[:,:2],
    edge_attr = np.array(edge_index)[:,-1],  # weight
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)




unique_nodes = set(g.edge_index.reshape(-1).tolist())

dict_map = {n:i for i,n in enumerate(unique_nodes)}

source, target = np.array(edge_index)[:,0], np.array(edge_index)[:,1]


edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    edge_attr = np.array(edge_index)[:,-1],
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (500,500)
}

graft.graph.plot_undirected_weighted(
    g,
    node_color=g.y,
    node_size=g.x,
    edge_weight_text=False,
    draw_options= dr_opts
)

---

int(4.503100e+18)

4503100000000000000

---

pdf! (node10) cc_num = 4.503100e+18

# import importlib


# # importlib을 사용하여 모듈 다시 로드
# importlib.reload(graft)

c_sample = 4.503100e+18

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)
edge_index = (np.array(edge_index)).astype(np.float64)
edge_index = np.array(edge_index)[np.array(edge_index)[:,0] != np.array(edge_index)[:,1]]
theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta*9)!= 1)*(np.exp(-edge_index[:,2]/theta*9))
edge_index = edge_index.tolist()
selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()
g = torch_geometric.data.Data(
    edge_index = np.array(edge_index)[:,:2],
    edge_attr = np.array(edge_index)[:,-1],  # weight
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)



unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = np.array(edge_index)[:,0], np.array(edge_index)[:,1]
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    edge_attr = np.array(edge_index)[:,-1],
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (150,150),
    'edge_marker_size': 1,
  'output':"ten-nodes.pdf",
}
graft.graph.plot_undirected_weighted(
    g,
    node_color=g.y,
    node_size=g.x,
    edge_weight_text=False,
    edge_weight_width_scale=3.0,
    draw_options= dr_opts,
)

pdf! (node10) cc_num = 4.809700e+12

c_sample = 4.809700e+12

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)
edge_index = (np.array(edge_index)).astype(np.float64)
edge_index = np.array(edge_index)[np.array(edge_index)[:,0] != np.array(edge_index)[:,1]]
theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta*9)!= 1)*(np.exp(-edge_index[:,2]/theta*9))
edge_index = edge_index.tolist()
selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()
g = torch_geometric.data.Data(
    edge_index = np.array(edge_index)[:,:2],
    edge_attr = np.array(edge_index)[:,-1],  # weight
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)



unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = np.array(edge_index)[:,0], np.array(edge_index)[:,1]
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    edge_attr = np.array(edge_index)[:,-1],
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (150,150),
    'edge_marker_size': 1,
  'output':"ten-nodes.pdf",
}
graft.graph.plot_undirected_weighted(
    g,
    node_color=g.y,
    node_size=g.x,
    edge_weight_text=False,
    edge_weight_width_scale=3.0,
    draw_options= dr_opts,
)

pdf! (node10) cc_num = 6.511350e+15

c_sample = 6.511350e+15

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)
edge_index = (np.array(edge_index)).astype(np.float64)
edge_index = np.array(edge_index)[np.array(edge_index)[:,0] != np.array(edge_index)[:,1]]
theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta*9)!= 1)*(np.exp(-edge_index[:,2]/theta*9))
edge_index = edge_index.tolist()
selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()
g = torch_geometric.data.Data(
    edge_index = np.array(edge_index)[:,:2],
    edge_attr = np.array(edge_index)[:,-1],  # weight
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)



unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = np.array(edge_index)[:,0], np.array(edge_index)[:,1]
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    edge_attr = np.array(edge_index)[:,-1],
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (150,150),
    'edge_marker_size': 1,
  'output':"ten-nodes.pdf",
}
graft.graph.plot_undirected_weighted(
    g,
    node_color=g.y,
    node_size=g.x,
    edge_weight_text=False,
    edge_weight_width_scale=3.0,
    draw_options= dr_opts,
)

pdf! (node10) cc_num = 5.020500e+11

c_sample = 5.020500e+11

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)
edge_index = (np.array(edge_index)).astype(np.float64)
edge_index = np.array(edge_index)[np.array(edge_index)[:,0] != np.array(edge_index)[:,1]]
theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta*9)!= 1)*(np.exp(-edge_index[:,2]/theta*9))
edge_index = edge_index.tolist()
selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()
g = torch_geometric.data.Data(
    edge_index = np.array(edge_index)[:,:2],
    edge_attr = np.array(edge_index)[:,-1],  # weight
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)



unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = np.array(edge_index)[:,0], np.array(edge_index)[:,1]
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    edge_attr = np.array(edge_index)[:,-1],
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (150,150),
    'edge_marker_size': 1,
  'output':"ten-nodes.pdf",
}
graft.graph.plot_undirected_weighted(
    g,
    node_color=g.y,
    node_size=g.x,
    edge_weight_text=False,
    edge_weight_width_scale=3.0,
    draw_options= dr_opts,
)

pdf하나더

# import importlib
# importlib.reload(graft)

c_sample = 2.720010e+15

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)
edge_index = (np.array(edge_index)).astype(np.float64)
edge_index = np.array(edge_index)[np.array(edge_index)[:,0] != np.array(edge_index)[:,1]]
theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta)!= 1)*(np.exp(-edge_index[:,2]/theta))
edge_index = edge_index.tolist()
selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()
g = torch_geometric.data.Data(
    edge_index = np.array(edge_index)[:,:2],
    edge_attr = np.array(edge_index)[:,-1],  # weight
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)



unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = np.array(edge_index)[:,0], np.array(edge_index)[:,1]
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    edge_attr = np.array(edge_index)[:,-1],
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (500,500),
    'edge_marker_size': 1,
    #'output':"ten-nodes.pdf",
}
graft.graph.plot_undirected_weighted(
    g,
    node_color=g.y,
    node_size=g.x,
    edge_weight_text=False,
    draw_options= dr_opts,
)

pdf?

---

# def graph_sample(df, c_sample):
#     def compute_time_difference(group):
#         n = len(group)
#         result = []
#         for i in range(n):
#             for j in range(n):
#                 time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
#                 result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
#         return result

#     groups = df[df['cc_num'] == c_sample]
#     edge_index = compute_time_difference(groups)
#     edge_index = (np.array(edge_index)).astype(np.float64)
#     edge_index = np.array(edge_index)[np.array(edge_index)[:,0] != np.array(edge_index)[:,1]]
#     theta = edge_index[:,2].mean()
#     edge_index[:,2] = (np.exp(-edge_index[:,2]/theta)!= 1)*(np.exp(-edge_index[:,2]/theta))

#     edge_index = edge_index.tolist()

#     selected_indices = df[df['cc_num'] ==  c_sample].index

#     g = data.Data(
#         edge_index=torch.tensor(edge_index)[:, :2],
#         edge_attr=torch.tensor(edge_index)[:, -1],
#         x=df.loc[selected_indices, 'amt'].tolist(),
#         y=df.loc[selected_indices, 'is_fraud'].tolist()
#     )

#     unique_nodes = set(g.edge_index.reshape(-1).tolist())
#     dict_map = {n: i for i, n in enumerate(unique_nodes)}
#     source, target = np.array(edge_index)[:, 0], np.array(edge_index)[:, 1]
#     edge_index_selected = torch.tensor([[dict_map[s] for s in source.tolist()], [dict_map[t] for t in target.tolist()]])

#     g = data.Data(
#         edge_index=edge_index_selected,
#         edge_attr=torch.tensor(edge_index)[:, -1],
#         x=df.loc[selected_indices, 'amt'].tolist(),
#         y=df.loc[selected_indices, 'is_fraud'].tolist()
#     )

#     dr_opts = {'output_size': (500, 500)}

#     plt.figure(figsize=dr_opts['output_size'])
#     graft.graph.plot_undirected_weighted(
#         g,
#         node_color=g.y,
#         node_size=g.x,
#         edge_weight_text=False,
#         draw_options=dr_opts
#     )

#     return plt

---

df50['cc_num'].value_counts()

cc_num
4.302480e+15    43
1.800650e+14    36
2.131740e+14    35
2.720430e+15    34
2.242540e+15    33
                ..
3.885950e+13     1
4.026220e+12     1
6.526450e+15     1
4.972230e+15     1
6.535330e+15     1
Name: count, Length: 932, dtype: int64

c_sample = 4.302480e+15

groups = df50[df50['cc_num'] == c_sample]
def compute_time_difference(group):
    n = len(group)
    result = []
    for i in range(n):
        for j in range(n):
            time_difference = abs(group.iloc[i].trans_date_trans_time.value - group.iloc[j].trans_date_trans_time.value)
            result.append([group.iloc[i].name, group.iloc[j].name, time_difference])
    return result

edge_index = compute_time_difference(groups)
edge_index = (np.array(edge_index)).astype(np.float64)
edge_index = np.array(edge_index)[np.array(edge_index)[:,0] != np.array(edge_index)[:,1]]
theta = edge_index[:,2].mean()
edge_index[:,2] = (np.exp(-edge_index[:,2]/theta)!= 1)*(np.exp(-edge_index[:,2]/theta))



edge_index = edge_index.tolist()
np.array(edge_index)[:,:2]  # edge!!
np.array(edge_index)[:,-1]  # weight
selected_indices1 = df50[df50['cc_num'] ==  c_sample].index
is_fraud_values1 = df50.loc[selected_indices1, 'is_fraud'].tolist()
x = df50.loc[selected_indices1, 'amt'].tolist()
y = df50.loc[selected_indices1, 'is_fraud'].tolist()

g = torch_geometric.data.Data(
    edge_index = np.array(edge_index)[:,:2],
    edge_attr = np.array(edge_index)[:,-1],  # weight
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)



unique_nodes = set(g.edge_index.reshape(-1).tolist())
dict_map = {n:i for i,n in enumerate(unique_nodes)}
source, target = np.array(edge_index)[:,0], np.array(edge_index)[:,1]
edge_index_selected2 = torch.tensor([[dict_map[s] for s in source.tolist()],[dict_map[t] for t in target.tolist()]])

g = torch_geometric.data.Data(
    edge_index = edge_index_selected2,
    edge_attr = np.array(edge_index)[:,-1],
    x = df50.loc[selected_indices1, 'amt'].tolist(),
    y = df50.loc[selected_indices1, 'is_fraud'].tolist()
)

dr_opts = { 
    'output_size': (500,500)
}

graft.graph.plot_undirected_weighted(
    g,
    node_color=g.y,
    node_size=g.x,
    edge_weight_text=False,
    draw_options= dr_opts
)

---

cc_num_counts = df50['cc_num'].value_counts()
cc_num_10_nodes = cc_num_counts[cc_num_counts == 20].index.tolist()

cc20 = df50[(df50['cc_num'].isin(cc_num_10_nodes))]

cc20.groupby("cc_num")['is_fraud'].sum().sort_values()

cc_num
3.576430e+15     2
2.131420e+14     3
4.377340e+15     7
4.681700e+12     7
2.233880e+15     7
6.761950e+11     8
2.720010e+15     8
1.800430e+14     8
4.149240e+15     9
4.393520e+12    10
3.770270e+14    10
6.762990e+11    11
6.554250e+15    11
2.291160e+15    12
4.223710e+18    12
4.601640e+12    12
4.198470e+12    12
5.714650e+11    12
3.533800e+15    13
3.003320e+13    13
4.048510e+15    14
5.020130e+11    14
6.042785e+10    14
Name: is_fraud, dtype: int64

c_sample = 4.958590e+18