帮我写一段GNN预测用户的借款意向程度的python代码

好的，下面是一个使用PyTorch和DGL实现的基于GNN的用户借款意向预测的Python代码示例：

import dgl
import torch
import torch.nn as nn
import torch.nn.functional as F
import dgl.function as fn

# 定义GCN模型
class GCN(nn.Module):
    def __init__(self, in_feats, hidden_feats, out_feats):
        super(GCN, self).__init__()
        self.conv1 = nn.Conv2d(in_feats, hidden_feats, kernel_size=1)
        self.conv2 = nn.Conv2d(hidden_feats, out_feats, kernel_size=1)

    def forward(self, g, features):
        h = self.conv1(features)
        h = F.relu(h)
        h = self.conv2(h)
        g.ndata['h'] = h
        g.update_all(message_func=fn.copy_u('h', 'm'),
                     reduce_func=fn.sum('m', 'h'))
        return g.ndata.pop('h')

# 构建图数据
user_data = {'age': [30, 35, 25, 28],
             'gender': ['M', 'F', 'F', 'M'],
             'education': ['Bachelor', 'Master', 'PhD', 'Bachelor'],
             'married': [False, True, False, True]}
loan_data = {'amount': [1000, 2000, 500, 1500],
             'term': [12, 6, 24, 12],
             'repayment': ['on-time', 'late', 'on-time', 'on-time']}
edge_data = {'borrow': [0, 1, 2, 3],
             'lend': [0, 1, 2, 3]}
user_g = dgl.graph(user_data, edge_data)
loan_g = dgl.graph(loan_data, edge_data)

# 构建节点特征
user_feats = torch.cat([F.one_hot(user_g.ndata['age']),
                        F.one_hot(user_g.ndata['gender']),
                        F.one_hot(user_g.ndata['education']),
                        F.one_hot(user_g.ndata['married'])], dim=-1)
loan_feats = torch.tensor(loan_data['amount'])

# 构建边特征
edge_feats = torch.tensor(loan_data['term'])

# 构建完整图
full_g = dgl.heterograph({('user', 'borrow', 'loan'): (user_g, loan_g)})

# 定义模型
gcn = GCN(user_feats.shape[-1], 128, 1)

# 训练模型
optimizer = torch.optim.Adam(gcn.parameters(), lr=0.01)
for epoch in range(10):
    logits = gcn(full_g, user_feats)
    loss = F.mse_loss(logits, loan_feats)
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()

# 预测新用户的借款意向
new_user_feats = torch.tensor([0, 1, 0, 0, 0, 0, 1, 0])
new_user_g = dgl.graph({'age': [26], 'gender': ['M'], 'education': ['Bachelor'], 'married': [True]})
new_loan_g = dgl.graph({'amount': [0], 'term': [0]})
new_full_g = dgl.heterograph({('user', 'borrow', 'loan'): (new_user_g, new_loan_g)})
with torch.no_grad():
    new_logit = gcn(new_full_g, new_user_feats)
    print('New user\'s loan intention:', new_logit.item())

需要注意的是，这只是一个简单的示例代码，实际应用中还需要进行更加细致的调整和优化。