data-analysis

name	Data Analysis
description	Analyze datasets by profiling, finding distributions and correlations, detecting anomalies, and translating statistics into actionable business insights.

Purpose

Turn raw data into actionable insight statements. Statistics are not insights. "Mean revenue is $4.2M" is a statistic. "Revenue grew 34% YoY but growth is decelerating" is an insight.

Steps

1. Profile the dataset: row count, column count, null percentages, type inference, duplicates
2. Compute descriptive statistics for every numeric column: mean, median, standard deviation, min/max, percentiles, skewness
3. Analyze distributions: identify normal, skewed, bimodal, or power-law shapes
4. Check correlations: flag strong (|r| > 0.7) and moderate (0.4-0.7) relationships
5. Segment by key categorical variables: compare group means, check for Simpson's Paradox
6. Detect anomalies: use IQR fences for skewed data, z-scores for normal data. Flag but never auto-remove.
7. Translate every finding into an insight statement using the DIKW framework
8. Create an Excel workbook with a summary sheet and supporting data sheets
9. Create a Word document with the top 5 insights, each following the insight template

Data Profiling Checklist

Metric	Red Flag
Null percentage	> 20% in key column = unreliable
Unique count = row count	Likely an ID column
Mixed types in column	Needs cleaning before analysis
> 1% duplicate rows	Dedup decision required
Mean >> median	Right-skewed, report median as "typical"

Insight Statement Template

For each finding, write:

• WHAT: [metric] is [value/behavior]