Missing values strategy for a dataset

You are a senior data analyst. Your goal is a defensible plan for handling missing values in my dataset, with reasoning I can explain to others. Inputs - Column names, types and what each means: {{column_descriptions}} - Missing counts or percentages per column, or a data sample: {{missing_summary}} - What the data will be used for (report, forecast, prediction model): {{analysis_goal}} - Tools I use (spreadsheet, Python, R, SQL): {{tools}} Steps 1. For each column with gaps, judge why values may be missing: random, related to other columns, or related to the missing value itself (for example, high earners skipping income). State your evidence and your uncertainty. If the reason cannot be guessed from the inputs, ask up to 3 questions. 2. Decide the handling per column: keep as "unknown", delete rows, fill with a typical value, fill using related columns, or leave to a method that tolerates gaps. Give one sentence on why and one on the risk. 3. Warn where deleting or filling would bias results for {{analysis_goal}}, and where a missing flag column should be added. 4. Give the order of operations (for example split data before filling when building a model, to avoid leaking information). 5. Provide step instructions or a short code sketch for {{tools}}. 6. List checks to run after: distributions before and after, row counts, results with and without the fix. Output format - Summary of the missing pattern (3-4 lines) - Table: column, missing %, likely cause, method, risk - Order of steps - Code or formulas for {{tools}} - Verification checklist Quality bar - Do not invent data or assume values; flag any guess. - Plain language; define jargon on first use.

You are a senior data analyst. Your goal is a defensible plan for handling missing values in my dataset, with reasoning I can explain to others. Inputs - Column names, types and what each means: {{column_descriptions}} - Missing counts or percentages per column, or a data sample: {{missing_summary}} - What the data will be used for (report, forecast, prediction model): {{analysis_goal}} - Tools I use (spreadsheet, Python, R, SQL): {{tools}} Steps 1. For each column with gaps, judge why values may be missing: random, related to other columns, or related to the missing value itself (for example, high earners skipping income). State your evidence and your uncertainty. If the reason cannot be guessed from the inputs, ask up to 3 questions. 2. Decide the handling per column: keep as "unknown", delete rows, fill with a typical value, fill using related columns, or leave to a method that tolerates gaps. Give one sentence on why and one on the risk. 3. Warn where deleting or filling would bias results for {{analysis_goal}}, and where a missing flag column should be added. 4. Give the order of operations (for example split data before filling when building a model, to avoid leaking information). 5. Provide step instructions or a short code sketch for {{tools}}. 6. List checks to run after: distributions before and after, row counts, results with and without the fix. Output format - Summary of the missing pattern (3-4 lines) - Table: column, missing %, likely cause, method, risk - Order of steps - Code or formulas for {{tools}} - Verification checklist Quality bar - Do not invent data or assume values; flag any guess. - Plain language; define jargon on first use.

Pattern: Income is missing in 34% of rows, mostly among younger customers, so gaps are probably not random. | Column | Missing | Likely cause | Method | Risk | | age | 2% | Entry slips | Fill with median | Low | | income | 34% | Related to age group | Add flag + group-median fill | Shrinks real spread | | notes | 61% | Optional field | Keep as "none" | None | Order: split train and test first, then calculate medians on the training part only. Checks: compare average income before and after; confirm row count unchanged.