Detecting Anomalies From Data | Dataford Interview Questions

Explain how you would detect anomalies from a dataset.

Implement detect_anomalies(data) for a non-empty numeric list with at least four values. Use the 1.5 × IQR rule, where quartiles are medians of the lower and upper halves after sorting, excluding the overall median when the length is odd. Return the original indices of values below Q1 - 1.5 * IQR or above Q3 + 1.5 * IQR, in ascending order.

Constraints

4 <= len(data) <= 10^5
Each value in data is an integer or floating-point number
The input contains finite numeric values
Use the median-of-halves quartile definition described in the prompt
Values exactly equal to a fence are not anomalies

Function Signature

def detect_anomalies(data):

Your question is Detecting Anomalies From Data. 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

Explain how you would detect anomalies from a dataset.

Constraints

4 <= len(data) <= 10^5
Each value in data is an integer or floating-point number
The input contains finite numeric values
Use the median-of-halves quartile definition described in the prompt
Values exactly equal to a fence are not anomalies

Function Signature

def detect_anomalies(data):

Your question is Detecting Anomalies From Data. 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.