I'm working on Ethereum Fraud detection dataset, 0 denoting normal and 1 denoting fradulent

I have train_account.csv as

transactions.csv as

I want to make a test_account.csv, where only accounts would be given and our task is to find if it is fradulent-1 or normal - 0

Rules I'm following to make below table

• If an account is present in train_account.csv account column then a flag column with value that is present for that account in train_account.csv will be added if not then I go for checking if that account is present in trasnaction['from_account'] or trasaction['to_account'] if yes then add flag value for that account in test_account to be 0 else 1

I'm planning to add a flag column based on the above rule, not to add or remove extra rows

PS: I'm beginner and have no clue to do this, Thanks in advance

I tried looping in columns but not sure how to check and add it to the result. some thing like this:

for i in test_account['account']:
    if i in train_account['account']:
        test_account[i]['flag'] = train_account[i]['flag']
    elif i in trasnaction['from_account'] or trasnaction['to_account']:
        test_account[i]['flag'] = 1
    else:
        test_account[i]['flag'] = 0

CodePudding user response：

Try this:

import numpy as np
# step 1
train_accounts = train_account['account'].values
test_account['flag'] = np.where(test_account['account'].isin(train_accounts), train_account['flag'], test_account['flag'])
# step 2
accounts = np.append(transactions['from_account'].values,transactions['to_account'].values)
test_account['flag'] = np.where(test_account['account'].isin(accounts), 1, 0)