The analyzer extracts every numeric and categorical signal from your captured JSONs and tests each against your mood, energy, and focus self-reports.

Numeric signals (temperature, pressure, AQI, DJIA close, BTC price, Kp index, moon illumination, etc.) get a Pearson correlation against each of mood, energy, focus. A coefficient near +1 means the signal rises with your score; near -1 means it falls; near 0 means no linear relationship.

Categorical signals (weather code, moon phase, drought classification, day of week, sun sign) get a group means analysis: average mood/energy/focus for entries in each category, with the within-category sample size shown.

Compound finder (the "rainy days when DJIA < 30,000" pattern) lets you specify up to two filter conditions and reports the average mood/energy/focus for the captures matching both, vs. the baseline across all captures.

Findings are ranked by absolute effect size weighted by sample size: a correlation of 0.8 over 6 entries is less trustworthy than 0.5 over 40 entries, and the ranker accounts for that.

Sample size matters. With fewer than 10 captures, almost any correlation is noise; the analyzer shows warnings when sample size is small. With 30+ captures, real patterns start to emerge. With 100+, you can probably trust what you see.

Privacy: all analysis runs in your browser. JSONs dropped here are read locally; nothing is uploaded.