Investment-Ready Countries by Call Duration

A telecom strategy team is evaluating where to expand premium network infrastructure.

Table: subscriber

• subscriber_id: unique user identifier
• subscriber_name: user display name
• phone_number: in format ccc-local_number, where ccc is a 3-digit country code

Table: country_code_map

• country_name: country display name
• country_code: unique 3-digit country code

Table: call_log

• caller_id: subscriber who initiated the call
• callee_id: subscriber who received the call
• duration_minutes: call duration in minutes
• Duplicate rows may exist and each row is a separate call event.
• caller_id != callee_id

Business metric definitions:

1. Global average duration:

   • Average duration_minutes across all rows in call_log.

2. Country endpoint average duration:

   • For every call row, evaluate caller and callee endpoints separately.
   • If an endpoint's subscriber phone code maps to a country in country_code_map, include that call duration once in that country's aggregate.
   • A single call may contribute:
     • once (only one mapped endpoint),
     • twice to two different countries,
     • or twice to the same country (both endpoints mapped there).

Task: Return countries whose endpoint average duration is strictly greater than the global average duration.

Output requirements:

• Return exactly one column named country.
• Sort result by country ascending.

Supported submission environments:

• SQL (SQLite 3.27.2)
• Pandas (Python)