Foundations and Explorer

Core rule: missing is not zero

Steady stores normal check-in values in a daily entry and repeatable activity as events. The analytics layer uses only values that are actually present.

State	Analytical meaning
key is missing	unknown, omitted
null or empty	not selected, omitted
0	actual numeric zero
no / false	deliberately negative status
na	not applicable, no signal

This distinction applies to Explorer, the Life Wheel and correlations. An unanswered day is interpreted neither as a bad day nor as a day without urges.

Data sources

flowchart LR
    E[entries] --> M[metric daily values]
    F[field definitions and versions] --> M
    X[events] --> M
    M --> EX[Explorer]
    M --> COR[Correlations]
    E --> RAD[Life Wheel]
    X --> RAD
    F --> RAD

Normal fields are combined through their logicalId. After a field type changes, older field versions remain analyzable; the version actually present on a given day is used.

Dynamic events additionally appear as virtual metrics:

• social connection,
• exercise duration and exertion,
• self-care duration, relaxation and clarity,
• urge intensity,
• trigger count,
• meal amount and balance.

Field-dependent analytics rules

The valid aggregation and chart type depend on the field type. Invalid or obsolete configurations are normalized.

Field type	Default aggregation	Default chart	Further aggregations
Boolean	mean	line	count, last value
Status	mean	line	count, last value
Scale	mean	line	count, last value
Duration	sum	bar	mean, count, last value
Number	mean	bar	sum, count, last value
Choice	count per answer	bar	no numeric mean
Free text	count of non-empty days	bar	text content is not interpreted numerically

auto resolves to the field type default or to the event aggregation. Exercise and self-care durations, for example, are summed; qualitative event values are averaged by default.

Explorer

Explorer is the direct time-series path. It first filters the selected range and then creates buckets by day, week, month, quarter or year.

flowchart TD
    S[selected metric] --> V[resolve field versions]
    V --> T{Field or event?}
    T -->|Field| A[collect entry values]
    T -->|Event| B[collect event values]
    A --> G[time buckets]
    B --> G
    G --> AG[type-aware aggregation]
    AG --> C[chart]
    AG --> D[optional breakdown]

Choice fields

Choice fields are not averaged through artificial numeric values. Every answer option is counted as a separate component and displayed as a distribution.

Breakdowns

A metric can be split by a categorical dimension, for example:

• exercise duration by activity,
• connection by person,
• a metric by status or choice field,
• events by meal, trigger or self-care activity.

The breakdown also calculates a descriptive association value as the share of between-group variance in total variance. It describes how strongly group means differ; it is neither proof of causality nor a significance test.