The interview report preserved the survey tables and the request to calculate response rate, but not the exact grouping or output contract. The following is a self-contained practice reconstruction of that task. You have two PostgreSQL tables: ```text survey ------ user_id BIGINT NOT NULL event_date DATE NOT NULL survey_event TEXT NOT NULL -- 'impression' or 'response' response INTEGER NULL -- 1 through 5 for a response; NULL for an impression users ----- user_id BIGINT PRIMARY KEY reg_date DATE NOT NULL country TEXT NOT NULL ``` Within a calendar date, a user has at most one `impression` row and at most one `response` row. Every `survey.user_id` has a matching `users` row. A response counts only when the same user also has an impression on the same date; ignore orphan response rows. All `response` rows contain an integer from 1 through 5. Write one PostgreSQL query that returns one row for each `event_date` and `country` with at least one survey impression. Return: - `survey_date` - `country` - `impressed_users`: number of distinct users with an impression that day - `responding_users`: number of those impressed users who also responded that day - `response_rate`: `responding_users / impressed_users` as a decimal between 0 and 1 Preserve numeric precision rather than formatting the rate as text or a percentage. Order the result by `survey_date` ascending and `country` ascending.