Sliding Window Anomaly Detection | Dataford Interview Questions

Given a stream of data, how would you implement a sliding window filter to detect anomalies?

Implement detect_anomalies(data, window_size, threshold). For each value, compare it with the mean of up to the previous window_size values. Return an array of zero-based indices where the absolute difference is greater than threshold; values without prior observations are never anomalous.

Constraints

1 <= data.length <= 5000
1 <= window_size <= 5000
Each data value is a finite real number
0 <= threshold
An observation with no prior values is never anomalous
An observation exactly threshold units from the mean is not anomalous

Function Signature

def detect_anomalies(data, window_size, threshold):

Your question is Sliding Window Anomaly Detection. 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 data, how would you implement a sliding window filter to detect anomalies?

Constraints

1 <= data.length <= 5000
1 <= window_size <= 5000
Each data value is a finite real number
0 <= threshold
An observation with no prior values is never anomalous
An observation exactly threshold units from the mean is not anomalous

Function Signature

def detect_anomalies(data, window_size, threshold):

Your question is Sliding Window Anomaly Detection. 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.