A chart prompt must state how missing, invalid, duplicate, and late records affect both marks and summary statistics. A missing temperature is not a zero-degree reading. Count eligible observations separately from scheduled slots, then name the denominator whenever a rate appears. Do not interpolate missing values unless the task explicitly authorizes a method and labels the result as estimated. When duplicate readings conflict, resolve the data issue before plotting rather than letting the model choose a convenient value. A chart can show uncertainty or a gap; it need not manufacture a continuous series.
Make this comfortable
Chart prompts: expose missing data and checked denominators
Operational case
C-47 has six scheduled slots and five valid observations. Two observed values exceed 5.0 degrees, so the observed exceedance rate is 2 divided by 5, or 40%; the missing slot is excluded from that denominator and remains visible as missing. The team does not call 40% the fraction of time above the limit. If the missing slot later receives a verified late reading, the chart version changes and the rate is recomputed. A conflicting duplicate at the same hour blocks release until the sensor record is reconciled.
Scheduled slots: 6; valid readings: 5; missing: 1.
Observed exceedances: 2; eligible denominator: 5.
Observed exceedance rate: 2 / 5 = 40%.
Missing slot: gap, not 0 °C.
Conflicting duplicate: hold chart until resolved.Performance and operating cost
A single pass over N records can validate slot membership and compute counts in O(N) time with O(U) storage for U unique slot keys. Reconciliation of duplicates may add review time, but it prevents a false aggregate from being repeated in reports. Displaying the denominator adds little space and removes ambiguity. Keep rate and count together so a viewer does not mistake a small sample for a stable operational trend.
Common Mistakes
- Do not plot a missing reading as zero.
- Do not divide by six when the stated rate is over five valid observations.
- Do not describe a reading fraction as elapsed exposure time.
Connected lessons
- Prompt engineering applications
- Prompt Engineering
- Aggregation prompts: pin the denominator and recompute the rate
- Dataset intake prompts: define a column before analyzing it
- Report summaries: preserve contrary results and missing data
- Chart prompts: state the decision and measured quantity
- Chart prompts: choose marks and scales by data type
- Chart prompts: produce a reproducible spec and data lineage
- Chart prompts: review rendered and nonvisual outputs
- Project: review a cold-chain temperature chart
- Chart-creation prompt decisions
prompt engineering
chart creation