Data Analytics for Accounting

Four analytics types

Descriptive (what happened) · Diagnostic (why) · Predictive (what will happen) · Prescriptive (what to do).

Process

Question → data → extract, transform, and load (ETL) → analyze → communicate.

Data

• Structured vs unstructured; nominal vs ordinal; primary and foreign keys.
• Quality: accuracy, completeness, consistency, timeliness, validity, uniqueness.

Statistics

• Mean (outlier-sensitive), median (robust), mode.
• Normal distribution: 68% / 95% / 99.7% within 1 / 2 / 3 standard deviations.
• Correlation −1 to +1, not causation. Regression y = a + bx; R² explanatory power; p < 0.05 significant.
• Benford's law: leading 1 ≈ 30%.

Charts

Trend → line · categories → bar · relationship → scatter · distribution → histogram · parts → stacked bar or pie.

Exam traps

• For skewed data such as salaries, the median describes the typical value better than the mean.
• Benford's law suits naturally occurring amounts spanning several orders of magnitude — not assigned numbers such as invoice numbers or prices set at $9.99.
• A primary key must be unique and never empty; a foreign key can repeat.
• A high R² shows fit, not that x causes y or that the model is right.