USE CASE — ANOMALY DETECTION

Know the moment something breaks.

Spot outliers in any time series. Tell a real incident apart from an expected seasonal swing.

  NO TRAINING REQUIRED   SUB-SECOND RESPONSE   PAY PER API CALL
DETECTED ANOMALIES · cpu-prod-01 DAILY · Z-SCORE
MON WED FRI SUN TUE
3
ANOMALIES FOUND
97%
DETECTION ACCURACY
5%
FALSE POSITIVES
97% DETECTION ACCURACY
5 MINUTES INTEGRATION TIME
99.99% UPTIME SLA

Built for signals worth acting on.

An alert nobody trusts is worse than no alert at all.

Catch it early

Alerts the moment a metric leaves the range it normally sits in.

REAL-TIME NOTIFICATIONS

Learns your patterns

Adapts to weekly and seasonal cycles instead of fixed thresholds.

ADAPTIVE THRESHOLDS

Fewer false alarms

Expected seasonal swings are filtered out, so alerts stay meaningful.

5% FALSE POSITIVE RATE
POST /v2/detect-anomalies

Built for your monitoring stack.

One endpoint. Send the series, get the outliers back.

forecastapi.com
curl -X POST https://forecastapi.com/v2/detect-anomalies \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "identifier": "cpu-prod-01",
    "frequency": "D",
    "data": [
      {"date": "2024-01-01", "value": 42},
      {"date": "2024-01-02", "value": 46},
      {"date": "2024-01-03", "value": 312}
    ],
    "anomaly_settings": {"method": "auto", "threshold": 3.0}
  }'

# Response
{
  "result": {
    "anomalies": [
      {
        "date": "2024-01-03",
        "value": 312,
        "type": "spike",
        "severity": "high",
        "z_score": 4.12,
        "expected_range": [38.2, 61.4],
        "deviation_percentage": 528.6
      }
    ],
    "analysis": {"method_used": "z_score", "confidence": "high"}
  },
  "meta": {"timing": {"total": 142.3}}
}

Detection and integrations included.

Four detection methods

auto, z-score, IQR, seasonal

Seasonal awareness

Weekly, monthly, and holiday cycles

Severity scoring

Spike or dip, ranked by deviation

Tunable sensitivity

Thresholds from 0.5 to 5.0

Streaming support

Detect as the data arrives

Monitoring tools

Datadog, New Relic, Grafana

Alerting

PagerDuty, Slack, Microsoft Teams

Time series stores

Prometheus, InfluxDB, Elasticsearch

Critical decisions, made easier.

Where catching it early changes the outcome.

Incident response

Catch the regression before the first support ticket.

Alert thresholds

Replace static limits with ranges learned from the data.

Capacity checks

Spot resource drift long before it saturates.

Revenue monitoring

Notice a conversion drop the day it starts.

Data quality

Flag broken pipelines when values go implausible.

Maintenance planning

Act on sensor drift before the equipment fails.

Trusted across industries.

From DevOps to business intelligence, anywhere a number can go wrong.

System & infrastructure

CPU, memory, and network watched for early warning signs.

Server performance monitoring
Database query anomalies
Network latency spikes

Business intelligence

Unusual movement in sales, engagement, and conversion.

Revenue anomaly detection
User engagement changes
Conversion rate drops

IoT & sensors

Temperature, vibration, and environmental data for predictive maintenance.

Equipment health monitoring
Environmental sensor alerts
Predictive maintenance

Pay only for what you use.

Scale your detection as you grow.

USAGE-BASED PRICING
First 1,000 calls/month Free
1,001 – 50,000 $0.0033/call
50,001 – 1,000,000 $0.0027/call
1,000,001+ $0.0016/call
Save 50% on every call with batch processing.
WHAT'S INCLUDED
Four detection methods
Seasonal-aware detection
Real-time streaming support
API & webhooks
Enterprise plans add dedicated support, custom SLAs, and on-prem deployment.

Catch it before it breaks.

Join the teams monitoring with ForecastAPI.
Your first 200 calls are free, no credit card required.