Agent Tools
series / anomaly
Runs three detectors over one series and reports what each saw: a robust z score on median and median absolute deviation, a robust seasonal-trend decomposition with a z score on the residual, and an isolation forest over value and step-to-previous. Every score is returned per point, so callers set their own threshold instead of trusting a verdict. The forest draws from a fixed seed: same request, same result. Under three full cycles the seasonal figure is marked as not robust.
POST https://tools.halowerk.com/v1/series/anomalyEingaben
| Feld | Typ | Pflicht | Beschreibung |
|---|---|---|---|
| points | array | Ja | Either plain numbers or objects with value and optionally time_unix_s (seconds since 1970), time (ISO 8601) and label. The service caps the whole request body at 256 KB, so long records reach that limit before maxItems. |
| methods | array | Nein | Which detectors to run. |
| robust_z_threshold | number | Nein | Absolute robust z score at which a point counts as an outlier. |
| season_period | integer | Nein | Points per season, e.g. 24 for hourly data with a daily rhythm. 0 leaves the seasonal part out. |
| stl_z_threshold | number | Nein | Threshold on the robust z score of the residual. |
| contamination | number | Nein | Expected share of outliers. Sets where the isolation forest cuts. |
| trees | integer | Nein | Number of isolation trees. |
| sample_size | integer | Nein | Points drawn per tree, capped at the series length. |
| seed | integer | Nein | Fixed starting value for the forest. Same seed, same answer. |
| min_votes | integer | Nein | How many detectors must agree before a point is called an anomaly. |
Vollständiges Eingabeschema anzeigen
{
"type": "object",
"required": [
"points"
],
"properties": {
"points": {
"type": "array",
"minItems": 3,
"maxItems": 5000,
"description": "Either plain numbers or objects with value and optionally time_unix_s (seconds since 1970), time (ISO 8601) and label. The service caps the whole request body at 256 KB, so long records reach that limit before maxItems."
},
"methods": {
"type": "array",
"minItems": 1,
"maxItems": 3,
"items": {
"type": "string",
"enum": [
"robust_z",
"stl_residual",
"isolation_forest"
]
},
"default": [
"robust_z",
"stl_residual",
"isolation_forest"
],
"description": "Which detectors to run."
},
"robust_z_threshold": {
"type": "number",
"minimum": 0.5,
"maximum": 20,
"default": 3.5,
"description": "Absolute robust z score at which a point counts as an outlier."
},
"season_period": {
"type": "integer",
"minimum": 0,
"maximum": 1000,
"default": 0,
"description": "Points per season, e.g. 24 for hourly data with a daily rhythm. 0 leaves the seasonal part out."
},
"stl_z_threshold": {
"type": "number",
"minimum": 0.5,
"maximum": 20,
"default": 3.5,
"description": "Threshold on the robust z score of the residual."
},
"contamination": {
"type": "number",
"minimum": 0.0001,
"maximum": 0.5,
"default": 0.02,
"description": "Expected share of outliers. Sets where the isolation forest cuts."
},
"trees": {
"type": "integer",
"minimum": 10,
"maximum": 300,
"default": 100,
"description": "Number of isolation trees."
},
"sample_size": {
"type": "integer",
"minimum": 8,
"maximum": 1024,
"default": 256,
"description": "Points drawn per tree, capped at the series length."
},
"seed": {
"type": "integer",
"minimum": 0,
"maximum": 2147483647,
"default": 42,
"description": "Fixed starting value for the forest. Same seed, same answer."
},
"min_votes": {
"type": "integer",
"minimum": 1,
"maximum": 3,
"default": 1,
"description": "How many detectors must agree before a point is called an anomaly."
}
}
}Veröffentlichtes Eingabebeispiel
{
"points": [
12.1,
12.4,
11.9,
12.3,
12,
12.2,
48.7,
12.1,
11.8,
12.5,
12.2,
12
],
"methods": [
"robust_z",
"isolation_forest"
],
"contamination": 0.1,
"seed": 42,
"min_votes": 2
}Ergebnis & Dokumentation
Maschinenlesbare Dokumentation dieses Hubs
Quelle dieser Seite: der veröffentlichte HALOWERK-Katalog. Stand: 2026-11-01, 15:19 UTC.
Weitere Endpunkte dieses Hubs
- Agent Tools0,002 USDC
abtest / evaluate
POSThttps://tools.halowerk.com/v1/abtest/evaluateReads one conversion experiment two ways at once. Frequentist: two-proportion test, p value, confidence interval on the absolute lift, and the sample size the requested minimum detectable effect needs at the requested power. Bayesian: Beta-
- Agent Tools0,002 USDC
address / validate
POSThttps://tools.halowerk.com/v1/address/validateParse, validate and normalize a supplied postal address. See the linked schema for inputs, coverage and limitations.
- Agent Tools0,004 USDC
agent / access
POSThttps://tools.halowerk.com/v1/agent/accessInspect published access rules for software agents on a web page. See the linked schema for inputs, coverage and limitations.