correlation_shift

Back to L2.5 | Browse all axes | Browse all options

Axis correlation_shift on sub-layer L2_5_C_correlation_shift (layer l2_5).

Sub-layer

L2_5_C_correlation_shift

Axis metadata

  • Default: 'none'

  • Sweepable: False

  • Status: operational

Operational status summary

  • Operational: 3 option(s)

  • Future: 0 option(s)

Options

delta_matrix – operational

Heatmap of (post - pre) correlation matrix.

L2.5.C correlation-shift view delta_matrix.

This option configures the correlation_shift axis on the L2_5_C_correlation_shift sub-layer of L2.5; output is emitted under manifest.diagnostics/l2_5/L2_5_C_correlation_shift/ alongside the other selected views.

When to use

Detecting cleaning steps that distort dependence structure; large entries flag pairs whose joint behaviour was altered.

References

  • macroforecast design Part 4: ‘diagnostic layers default-off; non-blocking; produce JSON + matplotlib views attached to manifest.diagnostics/.’

Related options: pre_post_overlay, none

Last reviewed 2026-05-05 by macroforecast author.

none – operational

Skip correlation-shift diagnostic.

L2.5.C correlation-shift view none.

This option configures the correlation_shift axis on the L2_5_C_correlation_shift sub-layer of L2.5; output is emitted under manifest.diagnostics/l2_5/L2_5_C_correlation_shift/ alongside the other selected views.

When to use

When L2 only changes scale (no transform that reshapes correlations).

References

  • macroforecast design Part 4: ‘diagnostic layers default-off; non-blocking; produce JSON + matplotlib views attached to manifest.diagnostics/.’

Related options: delta_matrix, pre_post_overlay

Last reviewed 2026-05-05 by macroforecast author.

pre_post_overlay – operational

Side-by-side pre / post correlation heatmaps.

L2.5.C correlation-shift view pre_post_overlay.

This option configures the correlation_shift axis on the L2_5_C_correlation_shift sub-layer of L2.5; output is emitted under manifest.diagnostics/l2_5/L2_5_C_correlation_shift/ alongside the other selected views.

When to use

Default visual; reveals magnitude and sign of changes simultaneously.

References

  • macroforecast design Part 4: ‘diagnostic layers default-off; non-blocking; produce JSON + matplotlib views attached to manifest.diagnostics/.’

Related options: delta_matrix, none

Last reviewed 2026-05-05 by macroforecast author.