I need to identify columns with missing values and calculate the percentage of missing data for each column in the dataset. Let me write a SQL query to check all columns in the raw.ai_job_dataset table for null values and calculate the missing percentage. ```json { "action": "tool_call", "reasoning": "I'll query the raw.ai_job_dataset table to count null values across all columns and calculate the percentage missing for each. This requires one comprehensive SELECT with COUNT(*) for total rows and COUNT(column_name) for non-null counts in each column, then computing the missing percentage.", "tool": "query_sql", "args": { "sql": "SELECT COUNT(*) as total_rows, COUNT(*) - COUNT(job_id) as job_id_nulls, ROUND(100.0 * (COUNT(*) - COUNT(job_id)) / COUNT(*), 2) as job_id_pct_missing, COUNT(*) - COUNT(job_title) as job_title_nulls, ROUND(100.0 * (COUNT(*) - COUNT(job_title)) / COUNT(*), 2) as job_title_pct_missing, COUNT(*) - COUNT(salary_usd) as salary_usd_nulls, ROUND(100.0 * (COUNT(*) - COUNT(salary_usd)) / COUNT(*), 2) as salary_usd_pct_missing, COUNT(*) - COUNT(salary_currency) as salary_currency_nulls, ROUND(100.0 * (COUNT(*) - COUNT(salary_currency)) / COUNT(*), 2) as salary_currency_pct_missing, COUNT(*) - COUNT(experience_level) as experience_level_nulls, ROUND(100.0 * (COUNT(*) - COUNT(experience_level)) / COUNT(*), 2) as experience_level_pct_missing, COUNT(*) - COUNT(employment_type) as employment_type_nulls, ROUND(100.0 * (COUNT(*) - COUNT(employment_type)) / COUNT(*), 2) as employment_type_pct_missing, COUNT(*) - COUNT(company_location) as company_location_nulls, ROUND(100.0 * (COUNT(*) - COUNT(company_location)) / COUNT(*), 2) as company_location_pct_missing, COUNT(*) - COUNT(company_size) as company_size_nulls, ROUND(100.0 * (COUNT(*) - COUNT(company_size)) / COUNT(*), 2) as company_size_pct_missing, COUNT(*) - COUNT(employee_residence) as employee_residence_nulls, ROUND(100.0 * (COUNT(*) - COUNT(employee_residence)) / COUNT(*), 2) as employee_residence_pct_missing, COUNT(*) - COUNT(remote_ratio) as remote_ratio_nulls, ROUND(100.0 * (COUNT(*) - COUNT(remote_ratio)) / COUNT(*), 2) as remote_ratio_pct_missing, COUNT(*) - COUNT(required_skills) as required_skills_nulls, ROUND(100.0 * (COUNT(*) - COUNT(required_skills)) / COUNT(*), 2) as required_skills_pct_missing, COUNT(*) - COUNT(education_required) as education_required_nulls, ROUND(100.0 * (COUNT(*) - COUNT(education_required)) / COUNT(*), 2) as education_required_pct_missing, COUNT(*) - COUNT(years_experience) as years_experience_nulls, ROUND(100.0 * (COUNT(*) - COUNT(years_experience)) / COUNT(*), 2) as years_experience_pct_missing, COUNT(*) - COUNT(industry) as industry_nulls, ROUND(100.0 * (COUNT(*) - COUNT(industry)) / COUNT(*), 2) as industry_pct_missing, COUNT(*) - COUNT(posting_date) as posting_date_nulls, ROUND(100.0 * (COUNT(*) - COUNT(posting_date)) / COUNT(*), 2) as posting_date_pct_missing, COUNT(*) - COUNT(application_deadline) as application_deadline_nulls, ROUND(100.0 * (COUNT(*) - COUNT(application_deadline)) / COUNT(*), 2) as application_deadline_pct_missing, COUNT(*) - COUNT(job_description_length) as job_description_length_nulls, ROUND(100.0 * (COUNT(*) - COUNT(job_description_length)) / COUNT(*), 2) as job_description_length_pct_missing, COUNT(*) - COUNT(benefits_score) as benefits_score_nulls, ROUND(100.0 * (COUNT(*) - COUNT(benefits_score)) / COUNT(*), 2) as benefits_score_pct_missing, COUNT(*) - COUNT(company_name) as company_name_nulls, ROUND(100.0 * (COUNT(*) - COUNT(company_name)) / COUNT(*), 2) as company_name_pct_missing FROM raw.ai_job_dataset", "purpose": "Count null values and calculate missing percentage for all columns in the dataset", "source": "raw" } } ```