Gnn connection with python

Hi ,

can anyone help me with the below error im facing : im currently trying the ethereum fraud detection github folder for GNN

While executing the gnn_fraud_detection file im getting an error in my .ipynb file in this section :
train_loader = conn.gds.neighborLoader(
v_in_feats=[“in_degree”,“out_degree”,“send_amount”,“send_min”,“recv_amount”,“recv_min”,“pagerank”],
v_out_labels=[“is_fraud”],
v_extra_feats=[“is_training”, “is_validation”],
output_format=“PyG”,
batch_size=hp[“batch_size”],
num_neighbors=hp[“num_neighbors”],
num_hops=hp[“num_hops”],
filter_by = “is_training”,
shuffle=True,
timeout=600000
)

This is the error im getting :
ConnectionError Traceback (most recent call last)
Input In [21], in <cell line: 1>()
----> 1 train_loader = conn.gds.neighborLoader(
2 v_in_feats=[“in_degree”,“out_degree”,“send_amount”,“send_min”,“recv_amount”,“recv_min”,“pagerank”],
3 v_out_labels=[“is_fraud”],
4 v_extra_feats=[“is_training”, “is_validation”],
5 output_format=“PyG”,
6 batch_size=hp[“batch_size”],
7 num_neighbors=hp[“num_neighbors”],
8 num_hops=hp[“num_hops”],
9 filter_by = “is_training”,
10 shuffle=True,
11 timeout=600000
12 )

Apologies for the non-standardize format, can someone help me as to why im facing this and what can be the possible solution.

@parkerrickson is the person who has written this i assume any leads on this would be helpful, many thanks

Hi @ms23 . Looks like there are a couple issues. First, my guess is that the database you are using does not have ML Workbench installed (download here: https://act.tigergraphlabs.com/). The second is that it appears you are using an old version of pyTigerGraph, you can upgrade with pip install -U pyTigerGraph.

1 Like