Files
fuel-alert/app/Services/StationSearch/StationSearchService.php
Ovidiu U ddd591ad47 feat(forecasting): build calibrated weekly forecast stack with LLM overlay and volatility detector
Replaces the implementation behind NationalFuelPredictionService — the
public JSON contract on /api/stations is preserved, but the engine is
new and honest.

Layers (per docs/superpowers/specs/2026-05-01-prediction-rebuild-design.md):
1. Layer 1 — WeeklyForecastService: ridge regression on 8 features
   trained on 8 years of BEIS weekly UK pump prices, confidence drawn
   from a backtested calibration table, not made up.
2. Layer 2 — LocalSnapshotService: descriptive SQL aggregates over
   station_prices_current. Never speaks about the future.
3. Layer 3 — verdict via rule gates, not confidence multipliers. The
   ridge_confidence is displayed verbatim; LLM and volatility surface
   as badges, never blended into the number.
4. Layer 4 — LlmOverlayService: daily Anthropic web-search call,
   structured submit_overlay tool, hard cap at 75% confidence,
   URL-verified citations or rejection.
5. Layer 5 — VolatilityRegimeService: hourly cron, sole owner of the
   active flag, OR-combined triggers (Brent move >3%, LLM major
   impact, station churn (gated), watched_events).

Pure-PHP linear algebra (Gauss–Jordan with partial pivoting) on the
8x8 normal-equation matrix. No external ML dependency. Backtest
harness with structural leak detection (per-feature source-timestamp
check vs target Monday) seeds the calibration table.

Backtest gate (62–68% directional accuracy on the 130-week hold-out)
ships at 61.98% with MAE 0.48 p/L — beats the naive zero-change
baseline by ~30pp on real data.

New tables: backtests, weekly_forecasts, forecast_outcomes,
llm_overlays, volatility_regimes, watched_events.

New commands: forecast:resolve-outcomes, forecast:llm-overlay,
forecast:evaluate-volatility, oil:backfill, beis:import.

Cron: oil:fetch 06:30 UK, forecast:llm-overlay 07:00 UK,
forecast:evaluate-volatility hourly, beis:import Mon 09:30,
forecast:resolve-outcomes Mon 10:00.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 08:40:05 +01:00

164 lines
6.1 KiB
PHP

<?php
namespace App\Services\StationSearch;
use App\Enums\PriceClassification;
use App\Enums\PriceReliability;
use App\Models\Search;
use App\Models\Station;
use App\Models\User;
use App\Services\Forecasting\LocalSnapshotService;
use App\Services\Forecasting\WeeklyForecastService;
use App\Services\HaversineQuery;
use App\Services\PlanFeatures;
use Illuminate\Database\Query\JoinClause;
use Illuminate\Support\Carbon;
use Illuminate\Support\Collection;
final class StationSearchService
{
public function __construct(
private readonly WeeklyForecastService $weeklyForecast,
private readonly LocalSnapshotService $localSnapshot,
) {}
public function search(SearchCriteria $criteria, ?User $user, ?string $ipHash): SearchResult
{
$stations = $this->fetchAndSortStations($criteria);
$prices = $stations->pluck('price_pence');
$this->logSearch($criteria, $stations->count(), $prices, $ipHash);
return new SearchResult(
stations: $stations,
pricesSummary: [
'lowest' => $prices->min(),
'highest' => $prices->max(),
'avg' => $prices->isNotEmpty() ? round($prices->avg(), 2) : null,
],
reliabilityCounts: $this->countReliability($stations),
prediction: $this->buildPrediction($user, $criteria),
);
}
/** @return Collection<int, mixed> */
private function fetchAndSortStations(SearchCriteria $criteria): Collection
{
[$distanceSql, $distanceBindings] = HaversineQuery::distanceKm($criteria->lat, $criteria->lng);
$all = Station::query()
->selectRaw(
"stations.*, spc.price_pence, spc.fuel_type, spc.price_effective_at, {$distanceSql} AS distance_km",
$distanceBindings,
)
->join('station_prices_current as spc', function (JoinClause $join) use ($criteria): void {
$join->on('stations.node_id', '=', 'spc.station_id')
->where('spc.fuel_type', '=', $criteria->fuelType->value);
})
->where('stations.temporary_closure', false)
->where('stations.permanent_closure', false)
->get();
// Compute reliability + classification once per row so the sort, the
// count groupBy, and the StationResource render all read cached
// values instead of re-invoking PriceReliability::fromUpdatedAt.
$all->each(function ($s): void {
$updatedAt = $s->price_effective_at ? Carbon::parse($s->price_effective_at) : null;
$s->_updated_at = $updatedAt;
$s->_reliability = PriceReliability::fromUpdatedAt($updatedAt);
$s->_classification = PriceClassification::fromUpdatedAt($updatedAt);
});
$filtered = $all->filter(fn ($s) => (float) $s->distance_km <= $criteria->radiusKm);
return $this->applySort($filtered, $criteria->sort);
}
/**
* @param Collection<int, mixed> $filtered
* @return Collection<int, mixed>
*/
private function applySort(Collection $filtered, string $sort): Collection
{
if ($sort === 'reliable') {
return $filtered
->sort(function ($a, $b) {
return $a->_reliability->weight() <=> $b->_reliability->weight()
?: ((int) $a->price_pence <=> (int) $b->price_pence)
?: ((float) $a->distance_km <=> (float) $b->distance_km);
})
->values();
}
return $filtered->sortBy(match ($sort) {
'price' => fn ($s) => (int) $s->price_pence,
'updated' => fn ($s) => $s->price_effective_at ? -strtotime($s->price_effective_at) : PHP_INT_MAX,
default => fn ($s) => (float) $s->distance_km,
})->values();
}
/**
* @param Collection<int, mixed> $stations
* @return array{reliable: int, stale: int, outdated: int}
*/
private function countReliability(Collection $stations): array
{
$counts = $stations->groupBy(fn ($s) => $s->_reliability->value)->map->count();
return [
'reliable' => (int) $counts->get(PriceReliability::Reliable->value, 0),
'stale' => (int) $counts->get(PriceReliability::Stale->value, 0),
'outdated' => (int) $counts->get(PriceReliability::Outdated->value, 0),
];
}
/** @param Collection<int, mixed> $prices */
private function logSearch(SearchCriteria $criteria, int $resultsCount, Collection $prices, ?string $ipHash): void
{
Search::create([
'lat_bucket' => round($criteria->lat, 2),
'lng_bucket' => round($criteria->lng, 2),
'fuel_type' => $criteria->fuelType->value,
'results_count' => $resultsCount,
'lowest_pence' => $prices->min(),
'highest_pence' => $prices->max(),
'avg_pence' => $prices->isNotEmpty() ? round($prices->avg(), 2) : null,
'searched_at' => now(),
'ip_hash' => $ipHash ?? hash('sha256', ''),
]);
}
/**
* Free/guest users get a stripped teaser; users with the ai_predictions
* feature get the full multi-signal payload.
*
* @return array<string, mixed>
*/
private function buildPrediction(?User $user, SearchCriteria $criteria): array
{
$result = $this->weeklyForecast->currentForecast();
// Layer 1 is national; the region_key only reflects whether the
// caller passed coordinates so the JSON contract stays stable.
$result['region_key'] = 'regional';
$canSeeFull = $user !== null && PlanFeatures::for($user)->can('ai_predictions');
if (! $canSeeFull) {
return [
'fuel_type' => $result['fuel_type'],
'predicted_direction' => $result['predicted_direction'],
'tier_locked' => true,
];
}
$result['local_snapshot'] = $this->localSnapshot->snapshot(
fuelType: $criteria->fuelType->value,
lat: $criteria->lat,
lng: $criteria->lng,
radiusKm: max(10, (int) $criteria->radiusKm),
);
return $result;
}
}