Detect Anomalies in Rolling Windows | Dataford Interview Questions

Given a stream of transaction data, identify anomalous patterns or duplicate entries within a rolling time window.

Implement detect_anomalies(transactions, window) for transactions ordered by nondecreasing timestamp. Each transaction is a dictionary with timestamp, transaction_id, user_id, and amount. Return {"duplicate_indices": [...], "anomalous_indices": [...]}. A transaction is a duplicate if its ID appeared within the previous window seconds. It is anomalous if it is the third or later transaction for the same user within that window. Indices refer to the original input list.

Constraints

1 <= len(transactions) <= 100000
0 <= transactions[i]["timestamp"] <= 10^9
0 <= window <= 10^9
Transactions are ordered by nondecreasing timestamp
Transaction IDs and user IDs are nonempty strings
Amounts are integers

Function Signature

def detect_anomalies(transactions, window):

Your question is Detect Anomalies in Rolling Windows. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem

Given a stream of transaction data, identify anomalous patterns or duplicate entries within a rolling time window.

Constraints

1 <= len(transactions) <= 100000
0 <= transactions[i]["timestamp"] <= 10^9
0 <= window <= 10^9
Transactions are ordered by nondecreasing timestamp
Transaction IDs and user IDs are nonempty strings
Amounts are integers

Function Signature

def detect_anomalies(transactions, window):

Your question is Detect Anomalies in Rolling Windows. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.