Ashley Allen
Dat 205 Project 3 Milestone
February 8, 2023
Client: La Banca Central Bank
Research Questions:
1. What behaviors are suggestive of fraud?
2. What kinds of financial transactions are frequently linked to fraud?
3. What links exist between the many fraudulent actions that have taken place?
Updated codebook
Description: The information was generated artificially. We were able to duplicate mobile
transactions using PaySim and a sample of actual transactions taken from a month's worth of
financial data. A worldwide business that participates in the global mobile banking network
provided the initial records. Currently, this organization offers services to more than 14 nations.
Components in a clean dataset:
1. (Integer) Step Actual units of measurement: An hour is the same as one step. We may search
for patterns of the frequency of fraudulent behavior as a function of time using the timestamp as
a variable.
2: (Varchar) Taking a closer look at elements like payment, transfer, debit, cash in, and cash out
can help us ascertain the nature of these transactions and find any relationships between
fraudulent activity and certain account types.
3. Amount: (Varchar) The transaction's total amount in local money. We can check for any links
between fraud and significant sums of money.
4. nameOrig: (Varchar) The client's name was identified. This variable is used to start the
transaction and determine whether the customer has a history of fraudulent behavior.
5. nameDest: (Varchar) The organization or individual receiving the transaction is named. This
gives us the opportunity to investigate fraud with this account.
6. isFraud: An integer We can find all links between the elements contributing to fraud by
utilizing the variable 1 or 0 to decide if the transaction was fraudulent or not.
7. isFlaggedFraud: (Integer) These variables mark any efforts to transmit more than $200,000 at
once as fraud (2) unflagged if none are flagged.
Justification: The data had all of the information needed to respond to our research's inquiries.
They are simple to grasp. Since the data does not contain account numbers or personally
identifying information, there should be no ethical concerns.
