Model Health

Diagnostics from your Meridian MCMC run — review model fit and convergence before interpreting results

All parameters converged

All R-hat values are below 1.1 — the MCMC chains have converged and results are reliable.

R-Squared

0.94

variance explained by model

MAPE

2.1%

mean absolute prediction error

WMAPE

1.8%

weighted prediction error

Model Fit: Expected vs Actual

Solid line is actual KPI, dashed is model prediction, shaded area is 90% credible interval

R-hat measures whether MCMC chains have converged — values below 1.1 indicate reliable posterior estimates

< 1.1 Converged1.1–1.2 Borderline> 1.2 Not converged

Media Coefficients

1.003

Adstock Alpha

1.008

Hill EC

1.012

Hill Slope

1.005

Control Coefficients

1.002

Time Effects (Knots)

1.018

Sigma (Noise)

1.001

How long each channel's effect persists after a spend event — higher carryover means longer-lasting impact

Higher carryover (slower decay) means the channel's effect persists longer after each spend event