<div class="section-label"><span class="num">01</span> Results</div>
<div class="section-title">What comes <em>back</em> from a prediction.</div>
Every successful <span class="mono">predict</span>, <span class="mono">predict_maxfit</span>, or <span class="mono">predict_grid</span> call returns a **PredictionResults** object. In thin clients (Python, R, and similar), sizes and core forecast fields sit on the **root**; deeper material is nested under named groups. Missing groups are simply absent (<span class="mono">None</span> in Python, <span class="mono">NULL</span> in R).
Settings that shape these fields live under [[Config/Overview|Config]]. The calls that produce them live under [[Functions/Overview|Functions]].
> [!warning]- C ABI / native integrators
> This chapter describes the **thin-client** layout (Python, R, and similar). The C ABI may group, name, or own fields differently: handles, copy-out APIs, retain flags, and nesting are not guaranteed to match these attribute paths. For bindings and native integration, treat the public header contract as authoritative: [[Install/Native Runtime|Native Runtime]] and the [C ABI repository](https://github.com/CambridgeSportsAnalytics/rbp-math-c-abi) (<span class="mono">rbp_math.h</span>).
<div class="pillars">
<div class="pillar">
<div class="pillar-num">↳ 01</div>
<div class="pillar-title">Always on</div>
<div class="pillar-text">Sizes, thresholds, yhat, fit, and related core measures on every successful run.</div>
</div>
<div class="pillar">
<div class="pillar-num">↳ 02</div>
<div class="pillar-title">Deeper detail</div>
<div class="pillar-text">prediction_weights, insights, auxiliary, and solo_distribution when the call stores them.</div>
</div>
<div class="pillar">
<div class="pillar-num">↳ 03</div>
<div class="pillar-title">Grid extras</div>
<div class="pillar-text">grid_insights on Grid runs; grid_cells only when retain_grid_objects lists keys.</div>
</div>
</div>
<p> </p>
<div class="section-label"><span class="num">02</span> How results are organized</div>
Thin-client layout (Python attribute access shown; R uses the same names with <span class="mono">
lt;/span>):
**Always present** (after a successful call)
```text
result PredictionResults
├── n_observations, n_variables, n_thresholds
├── thresholds
└── yhat, fit, adjusted_fit, agreement, asymmetry, k_fit, outlier_influence
(+ yhat_linear when include_linear_regression was on)
```
**Often present** on Predict / MaxFit (may be missing on Grid)
```text
result
├── prediction_weights include, weights, weights_excluded
├── insights relevance, similarity, info_x, info_theta, …
├── auxiliary phi, lambda_sq, full_var, …
└── solo_distribution y_solo, xi_solo, statistics, distribution
```
**Path-specific**
```text
result
├── maxfit_index MaxFit only
├── grid_insights usual on Grid (no retain needed)
│ ├── variable_weights
│ ├── impact_on_fit also: marginal_contribution_to_conviction
│ ├── impact_on_prediction also: marginal_contribution_to_prediction
│ ├── component_contribution_to_prediction
│ └── xi_solo_composite
└── grid_cells only if retain_grid_objects lists keys
├── k_cells, combi_cells, ysolo_cells
├── relevance yhat_cells, adjusted_fit_cells, …
└── similarity same cell fields for similarity censor
```
| Layer | Attribute / group | When you see it |
| --- | --- | --- |
| Core (root) | <span class="mono">yhat</span>, <span class="mono">fit</span>, … | Every successful call |
| Weights | <span class="mono">prediction_weights</span> | Common on Predict / MaxFit |
| Scores | <span class="mono">insights</span> | Common on Predict / MaxFit |
| Auxiliary | <span class="mono">auxiliary</span> | Common on Predict / MaxFit |
| Solo | <span class="mono">solo_distribution</span> | Common on Predict / MaxFit |
| MaxFit marker | <span class="mono">maxfit_index</span> | MaxFit |
| Grid insights | <span class="mono">grid_insights</span> | Usual on Grid |
| Grid cells | <span class="mono">grid_cells</span> | Only if <span class="mono">retain_grid_objects</span> is set |
<div class="solution-list">
<div class="solution-row">
<div class="solution-num">/01</div>
<div class="solution-title"><a href="/Results/Core">Core</a></div>
<div class="solution-desc">Root sizes, thresholds, yhat, fit, and related always-on fields.</div>
</div>
<div class="solution-row">
<div class="solution-num">/02</div>
<div class="solution-title"><a href="/Results/Weights%20&%20Insights">Weights & Insights</a></div>
<div class="solution-desc">prediction_weights and insights groups.</div>
</div>
<div class="solution-row">
<div class="solution-num">/03</div>
<div class="solution-title"><a href="/Results/Auxiliary">Auxiliary</a></div>
<div class="solution-desc">Per-threshold diagnostics under auxiliary.</div>
</div>
<div class="solution-row">
<div class="solution-num">/04</div>
<div class="solution-title"><a href="/Results/Solo%20Distribution">Solo Distribution</a></div>
<div class="solution-desc">solo_distribution: vectors, statistics, optional histogram.</div>
</div>
<div class="solution-row">
<div class="solution-num">/05</div>
<div class="solution-title"><a href="/Results/Grid%20Insights">Grid Insights</a></div>
<div class="solution-desc">grid_insights: Impact on Fit, Impact on Prediction, variable weights.</div>
</div>
<div class="solution-row">
<div class="solution-num">/06</div>
<div class="solution-title"><a href="/Results/Grid%20Cells">Grid Cells</a></div>
<div class="solution-desc">grid_cells when you set retain_grid_objects.</div>
</div>
</div>
## Sizes at a glance
| Symbol | Meaning | Root field |
| --- | --- | --- |
| N | Observations | <span class="mono">n_observations</span> |
| K | Variables | <span class="mono">n_variables</span> |
| T | Thresholds on this result | <span class="mono">n_thresholds</span> |
| Q | Attribute combinations (Grid cells) | inferred from retained cell arrays |
- **Predict:** T matches the thresholds you passed
- **MaxFit:** usually T = 1 (the winner)
- **Grid:** usually T = 1 (the composite); scalar forecast: first element of <span class="mono">yhat</span>
> [!tip]
> Check that a nested group exists before you use it. In Python it may be <span class="mono">None</span>; in R it may be <span class="mono">NULL</span>.
<div class="btn-row">
<a class="btn-primary" href="/Results/Core">Read core fields</a>
<a class="btn-ghost" href="/Functions/Overview">Back to Functions</a>
</div>