PROBLEMS WORKING WITH OWN DATASET (GCn)

[Akashkalakonda]

e_x

source target
0 0.0 1.0
1 0.0 2.0
2 0.0 3.0
3 0.0 4.0
4 0.0 5.0
5 0.0 6.0
6 0.0 7.0
7 0.0 8.0
8 0.0 8.5
9 0.0 9.0

e_x=torch.tensor(e_x.values,dtype=torch.long)
e_x
tensor([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 2, 3, 4, 5, 6, 7, 8, 8, 9]])

x_h=pd.DataFrame(([0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000], [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000] , [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000] , [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000] , [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000], [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000] , [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000], [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000] , [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000], [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0174, 0.0000, 0.0248,
0.0000] ))

x_h= torch.tensor(x_h.values, dtype=torch.long)

from torch_geometric.data import Data
data = Data(x=x_h, edge_index=e_x)

import torch
from torch.nn import Linear
from torch_geometric.nn import GCNConv

class GCN(torch.nn.Module):
def init(self):
super().init()
torch.manual_seed(1234)
self.conv1 = GCNConv(10, 4)
self.conv2 = GCNConv(4, 4)
self.conv3 = GCNConv(4, 2)
self.classifier = Linear(2, 2)

def forward(self, x, edge_index):
    h = self.conv1(x, edge_index)
    h = h.tanh()
    h = self.conv2(h, edge_index)
    h = h.tanh()
    h = self.conv3(h, edge_index)
    h = h.tanh()  # Final GNN embedding space.
    
    # Apply a final (linear) classifier.
    out = self.classifier(h)

model = GCN()
print(model)
GCN(
(conv1): GCNConv(10, 4)
(conv2): GCNConv(4, 4)
(conv3): GCNConv(4, 2)
(classifier): Linear(in_features=2, out_features=2, bias=True)
)

from IPython.display import Javascript # Restrict height of output cell.
display(Javascript('''google.colab.output.setIframeHeight(0, true, {maxHeight: 300})'''))

model = GCN()
optimizer = torch.optim.Adam(model.parameters(), lr=0.01, weight_decay=5e-4)
criterion = torch.nn.CrossEntropyLoss()

def train():
model.train()
optimizer.zero_grad() # Clear gradients.
out = model(data.x, data.edge_index) # Perform a single forward pass.
loss = criterion(out[data.train_mask], data.y) # Compute the loss solely based on the training nodes.
loss.backward() # Derive gradients.
optimizer.step() # Update parameters based on gradients.
return loss

def test():
model.eval()
out = model(data.x, data.edge_index)
pred = out.argmax(dim=1) # Use the class with highest probability.
test_correct = pred[data.test_mask] == data.y[data.test_mask] # Check against ground-truth labels.
test_acc = int(test_correct.sum()) / int(data.test_mask.sum()) # Derive ratio of correct predictions.
return test_acc

for epoch in range(1, 101):
loss = train()
print(f'Epoch: {epoch:03d}, Loss: {loss:.4f}')

AND THE ERRORS I GET ARE:

RuntimeError Traceback (most recent call last)
in
25
26 for epoch in range(1, 101):
---> 27 loss = train()
28 print(f'Epoch: {epoch:03d}, Loss: {loss:.4f}')

6 frames
in train()
9 model.train()
10 optimizer.zero_grad() # Clear gradients.
---> 11 out = model(data.x, data.edge_index) # Perform a single forward pass.
12 loss = criterion(out[data.train_mask], data.y) # Compute the loss solely based on the training nodes.
13 loss.backward() # Derive gradients.

/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
1128 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
1129 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1130 return forward_call(*input, **kwargs)
1131 # Do not call functions when jit is used
1132 full_backward_hooks, non_full_backward_hooks = [], []

in forward(self, x, edge_index)
14
15 def forward(self, x, edge_index):
---> 16 h = self.conv1(x, edge_index)
17 h = h.tanh()
18 h = self.conv2(h, edge_index)

/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
1128 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
1129 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1130 return forward_call(*input, **kwargs)
1131 # Do not call functions when jit is used
1132 full_backward_hooks, non_full_backward_hooks = [], []

/usr/local/lib/python3.7/dist-packages/torch_geometric/nn/conv/gcn_conv.py in forward(self, x, edge_index, edge_weight)
192 edge_index = cache
193
--> 194 x = self.lin(x)
195
196 # propagate_type: (x: Tensor, edge_weight: OptTensor)

/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
1128 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
1129 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1130 return forward_call(*input, **kwargs)
1131 # Do not call functions when jit is used
1132 full_backward_hooks, non_full_backward_hooks = [], []

/usr/local/lib/python3.7/dist-packages/torch_geometric/nn/dense/linear.py in forward(self, x)
116 x (Tensor): The features.
117 """
--> 118 return F.linear(x, self.weight, self.bias)
119
120 @torch.no_grad()

RuntimeError: expected scalar type Long but found Float

I AM COMPLETELY STRUCK HERE. AND MY COMMUNITY PLEASE HELP ME OUT WITH SOLUTION

[Akashkalakonda]

@rusty1s Can u help me out with issues

[EdisonLeeeee]

x_h should be float instead of long:

x_h = torch.tensor(x_h.values, dtype=torch.float)

[Akashkalakonda]

thanks sir.....but the problem hasnt solved yet
there are more errors:

---

KeyError Traceback (most recent call last)
/usr/local/lib/python3.7/dist-packages/torch_geometric/data/storage.py in getattr(self, key)
60 try:
---> 61 return self[key]
62 except KeyError:

4 frames
KeyError: 'train_mask'

During handling of the above exception, another exception occurred:

AttributeError Traceback (most recent call last)
/usr/local/lib/python3.7/dist-packages/torch_geometric/data/storage.py in getattr(self, key)
62 except KeyError:
63 raise AttributeError(
---> 64 f"'{self.class.name}' object has no attribute '{key}'")
65
66 def setattr(self, key: str, value: Any):

AttributeError: 'GlobalStorage' object has no attribute 'train_mask'

[EdisonLeeeee]

You were doing a node classification task without providing supervision data.y and train_mask/val_mask/test_mask that specify the training/validation/test nodes. There is only a graph with x and edge_index provided.

[Akashkalakonda]

y_label=pd.DataFrame([1,1,0,0,1,1,1,0,0,1])
y=y_label.to_numpy()
y.shape
(10, 1)

from torch_geometric.data import Data
data = Data(x=x_h, edge_index=e_x , y=y)

---

KeyError Traceback (most recent call last)
/usr/local/lib/python3.7/dist-packages/torch_geometric/data/storage.py in getattr(self, key)
60 try:
---> 61 return self[key]
62 except KeyError:

4 frames
/usr/local/lib/python3.7/dist-packages/torch_geometric/data/storage.py in getitem(self, key)
80 def getitem(self, key: str) -> Any:
---> 81 return self._mapping[key]
82

KeyError: 'train_mask'

During handling of the above exception, another exception occurred:

AttributeError Traceback (most recent call last)
in
25
26 for epoch in range(1, 101):
---> 27 loss = train()
28 print(f'Epoch: {epoch:03d}, Loss: {loss:.4f}')

in train()
10 optimizer.zero_grad() # Clear gradients.
11 out = model(data.x, data.edge_index) # Perform a single forward pass.
---> 12 loss = criterion(out[data.train_mask], data.y) # Compute the loss solely based on the training nodes.
13 loss.backward() # Derive gradients.
14 optimizer.step() # Update parameters based on gradients.

/usr/local/lib/python3.7/dist-packages/torch_geometric/data/data.py in getattr(self, key)
426 "dataset, remove the 'processed/' directory in the dataset's "
427 "root folder and try again.")
--> 428 return getattr(self._store, key)
429
430 def setattr(self, key: str, value: Any):

/usr/local/lib/python3.7/dist-packages/torch_geometric/data/storage.py in getattr(self, key)
62 except KeyError:
63 raise AttributeError(
---> 64 f"'{self.class.name}' object has no attribute '{key}'")
65
66 def setattr(self, key: str, value: Any):

AttributeError: 'GlobalStorage' object has no attribute 'train_mask'

????

[EdisonLeeeee]

You also need to define your train_mask/val_mask/test_mask in advance. For example (in case you have 10 nodes):

data.train_mask = torch.tensor([1, 1, 1, 0, 0, 0, 0, 0, 0, 0], dtype=torch.bool)
data.val_mask = torch.tensor([0, 0, 0, 1, 1, 0, 0, 0, 0, 0], dtype=torch.bool)
data.test_mask = torch.tensor([0, 0, 0, 0, 0, 1, 1, 1, 1, 1], dtype=torch.bool)

Also, y should be a 1D Tensor with shape [10] instead of [10, 1].

[Akashkalakonda]

more errors sir:

KeyError Traceback (most recent call last)
/usr/local/lib/python3.7/dist-packages/torch_geometric/data/storage.py in getattr(self, key)
60 try:
---> 61 return self[key]
62 except KeyError:

4 frames
KeyError: 'train_mask'

During handling of the above exception, another exception occurred:

AttributeError Traceback (most recent call last)
/usr/local/lib/python3.7/dist-packages/torch_geometric/data/storage.py in getattr(self, key)
62 except KeyError:
63 raise AttributeError(
---> 64 f"'{self.class.name}' object has no attribute '{key}'")
65
66 def setattr(self, key: str, value: Any):

AttributeError: 'GlobalStorage' object has no attribute 'train_mask'

ANY SOLUTIONS?

[LeoGori]

It seems that you have not extended your data with train_mask attribute, you can generate it through the definition of a RandomNodeSplit object, e.g.:

from torch_geometric.transforms import RandomNodeSplit
rns = RandomNodeSplit()
data = rns(data)

In this way, you are generating random train_mask, validation_mask and test_mask attributes on your data object

[Akashkalakonda]

hello sir @EdisonLeeeee
need ur help sir

[Akashkalakonda]

import pandas as pd
import numpy as np
import torch
from torch_geometric.data import Data
from sklearn.preprocessing import LabelEncoder
import torch_geometric.transforms as T
from torch_geometric.nn import GCNConv
import torch.nn.functional as F

Load the data

x = pd.read_csv('tempdataset.csv', index_col=0)
edge_index = pd.read_csv('tempdatasetsourceandtarget.csv', index_col=0)
y = pd.read_excel('y.xlsx')

Drop unnecessary columns from x

x = x.drop(columns=['ID', 'Name', 'Screen Name', 'Location'])

Convert data to numpy arrays

x = x.to_numpy()
edge_index = edge_index.to_numpy()
y = y.to_numpy().squeeze()

Encode edge_index columns as integers

le = LabelEncoder()
row, col = edge_index.T
le.fit(np.concatenate((row, col)))
row = le.transform(row)
col = le.transform(col)

Convert data to tensors

x = torch.tensor(x, dtype=torch.float)
edge_index = torch.tensor(np.stack([row, col], axis=0))
y = torch.tensor(y)

Create a PyTorch Geometric Data object

data = Data(x=x, edge_index=edge_index, y=y)
#Data(x=[114, 4], edge_index=[2, 114], y=[114])

Split the data into train, val, and test sets

data = T.RandomNodeSplit(num_val=0.2, num_test=0.4)(data)

Set device

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

Define the GCN model

class GCN(torch.nn.Module):
def init(self, in_channels, hidden_channels, out_channels):
super().init()
self.conv1 = GCNConv(in_channels, hidden_channels, cached=True)
self.conv2 = GCNConv(hidden_channels, out_channels, cached=True)

def forward(self, x, edge_index, edge_weight=None):
    x = F.dropout(x, p=0.5, training=self.training)
    x = self.conv1(x, edge_index, edge_weight).relu()
    x = F.dropout(x, p=0.5, training=self.training)
    x = self.conv2(x, edge_index, edge_weight)
    return x

Move model and data to the designated device

model = GCN(data.x.size(-1), 64, data.y.max().item()+1)
model, data = model.to(device), data.to(device)

Set up the optimizer

optimizer = torch.optim.Adam(model.parameters(), lr=0.01)

Define the training and evaluation functions

def train():
model.train()
optimizer.zero_grad()
out = model(data.x, data.edge_index, data.edge_weight)
loss = F.cross_entropy(out[data.train_mask], data.y[data.train_mask])
loss.backward()
optimizer.step()
return float(loss)

@torch.no_grad()
def test():
model.eval()
pred = model(data.x, data.edge_index, data.edge_weight).argmax(dim=-1)

accs = []
for mask in [data.train_mask, data.val_mask, data.test_mask]:
    accs.append(int((pred[mask] == data.y[mask]).sum()) / int(mask.sum()))
return accs

Train and evaluate the model

for epoch in range(1, 101):
loss = train()
accs = test()
print(f'Epoch {epoch}: loss = {loss:.4f}, acc = {accs[0]:.4f}, val_acc = {accs[1]:.4f}, test_acc = {accs[2]:.4f}')

this is my code
and this is the error i am not able to solve
RuntimeError: index 114 is out of bounds for dimension 0 with size 114
any suggestions here? @EdisonLeeeee and @LeoGori

[rusty1s]

What does data.validate() return? It looks like edge_index.max() >= data.num_nodes, which is not allowed.

[Akashkalakonda]

i didnt get it sir @rusty1s . is there any problem with the dataset i am using

[Akashkalakonda]

import pandas as pd
import numpy as np
import torch
from torch_geometric.data import Data
from sklearn.preprocessing import LabelEncoder
import torch_geometric.transforms as T
from torch_geometric.nn import GCNConv
import torch.nn.functional as F

Load the data

x = pd.read_csv('tempdataset.csv', index_col=0)
edge_index = pd.read_csv('tempdatasetsourceandtarget.csv', index_col=0)
y = pd.read_excel('y.xlsx')

Drop unnecessary columns from x

x = x.drop(columns=['ID', 'Name', 'Screen Name', 'Location'])

Convert data to numpy arrays

x = x.to_numpy()
edge_index = edge_index.to_numpy()
y = y.to_numpy().squeeze()

Encode edge_index columns as integers

le = LabelEncoder()
row, col = edge_index.T
le.fit(np.concatenate((row, col)))
row = le.transform(row)
col = le.transform(col)

Convert data to tensors

x = torch.tensor(x, dtype=torch.float)
edge_index = torch.tensor(np.stack([row, col], axis=0))
y = torch.tensor(y)

Create a PyTorch Geometric Data object

data = Data(x=x, edge_index=edge_index, y=y)

Split the data into train, val, and test sets

data = T.RandomNodeSplit(num_val=0.2, num_test=0.4)(data)

Set device

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

Define the GCN model

class GCN(torch.nn.Module):
def init(self, in_channels, hidden_channels, out_channels):
super().init()
self.conv1 = GCNConv(in_channels, hidden_channels, cached=True)
self.conv2 = GCNConv(hidden_channels, out_channels, cached=True)

def forward(self, x, edge_index, edge_weight=None):
    x = F.dropout(x, p=0.5, training=self.training)
    x = self.conv1(x, edge_index, edge_weight).relu()
    x = F.dropout(x, p=0.5, training=self.training)
    x = self.conv2(x, edge_index, edge_weight)
    return x

Move model and data to the designated device

model = GCN(data.x.size(-1), 64, data.y.max().item()+1)
model, data = model.to(device), data.to(device)

Set up the optimizer

optimizer = torch.optim.Adam(model.parameters(), lr=0.01)

Define the training and evaluation functions

def train():
model.train()
optimizer.zero_grad()
out = model(data.x, data.edge_index, data.edge_weight)
loss = F.cross_entropy(out[data.train_mask], data.y[data.train_mask])
loss.backward()
optimizer.step()
return float(loss)

@torch.no_grad()
def test():
model.eval()
pred = model(data.x, data.edge_index, data.edge_weight).argmax(dim=-1)

accs = []
for mask in [data.train_mask, data.val_mask, data.test_mask]:
    accs.append(int((pred[mask] == data.y[mask]).sum()) / int(mask.sum()))
return accs

Train and evaluate the model

for epoch in range(1, 101):
loss = train()
accs = test()
print(f'Epoch {epoch}: loss = {loss:.4f}, acc = {accs[0]:.4f}, val_acc = {accs[1]:.4f}, test_acc = {accs[2]:.4f}')

this is the entire code ihave been using . actually i am trying to perform node classification using GCN. i am using a custom dataset. i made a twitter dataset, and my goal is bot detection. the three data files which are given as input in the code are : x, edge_index, y. x has features of users like Name, Screen Name, Friends , Followers, Favorites, Location, Tweets, ID. 'edge_index' describes the link or to be precise edges. finally 'y' is label which specifies whether the user is human or bot ( 1 for bot and 0 for human). this is all about my dataset. for some reason i am not able to execute the code.
tempdataset_1.csv
tempdatasetsourceandtarget_1.csv
y.xlsx
above are the dummy datasets i have been using.

ERROR: RuntimeError: index 114 is out of bounds for dimension 0 with size 114
i am struck here.

[rusty1s]

Yes, the issue is that x.size() == (114, 4) and edge_index.max() == 114. As such, indexing with edge_index into x will raise an index error. The edge_index needs to be in interval 0 to x.size(0) - 1 == 113.

[Akashkalakonda]

thank you sir. i solved the error. you are a life savor

[ahsantfw]

Hi @AnotherAvenger, I have this dataset file and I want to train a GCN with 1 hidden layer only, Can you help me out how to load the data and give it to this current model here

This is how it looks like