Skip to content
CloudPi Docs

Cost Forecasting

Overview

CloudPi Cost Forecasting uses advanced machine learning algorithms to predict future cloud spending based on historical usage patterns, seasonal trends, and growth trajectories. Make informed decisions with accurate cost projections.

Key Features

  • AI-Powered Predictions: Machine learning models trained on your historical data
  • Multi-Cloud Forecasting: Unified forecasts across AWS, Azure, and GCP
  • Confidence Intervals: View forecast ranges with statistical confidence levels
  • Scenario Analysis: Model different growth scenarios
  • Anomaly Detection: Identify unusual spending patterns
  • Forecast Accuracy Tracking: Monitor prediction accuracy over time

Viewing Forecasts

Accessing Forecast Dashboard

  1. Navigate to Cost Management > Forecast
  2. Select the time range for forecast (30, 60, 90 days, or custom)
  3. Choose forecast scope:
  4. All Clouds: Aggregate forecast across providers
  5. By Provider: Separate forecasts for each cloud provider
  6. By Service: Service-level cost predictions
  7. By Tag: Forecast by cost allocation tags

Understanding Forecast Visualization

The forecast dashboard displays:

  • Historical Data: Past 90 days of actual spending
  • Forecast Line: Predicted spending trajectory
  • Confidence Bands: Upper and lower prediction bounds (80% confidence)
  • Budget Comparison: Forecast vs. existing budgets
  • Key Metrics: Total forecasted spend, growth rate, potential overruns

Forecast Models

Linear Forecast

Best for stable, predictable workloads with consistent growth.

Use Case: Production environments with steady usage

Seasonal Forecast

Accounts for recurring patterns (weekly, monthly, quarterly cycles).

Use Case: Retail applications with seasonal traffic

Growth-Adjusted Forecast

Incorporates business growth metrics and expansion plans.

Use Case: Scaling startups or new product launches

Custom Forecast

Define your own parameters and growth assumptions.

Use Case: Unique business models or major infrastructure changes

Scenario Analysis

Model different future scenarios:

Optimistic Scenario

  • Lower growth rate
  • Increased optimization adoption
  • Reserved instance utilization

Expected Scenario

  • Current trajectory continues
  • Moderate optimization efforts
  • Baseline forecast

Pessimistic Scenario

  • Higher growth rate
  • Minimal optimization
  • Conservative planning

Forecast Configuration

Adjusting Forecast Parameters

  1. Go to Forecast Settings
  2. Configure:
  3. Historical Window: Data range for model training (3-12 months)
  4. Seasonality: Enable/disable seasonal adjustments
  5. Growth Rate: Manual growth rate override
  6. Exclusions: Exclude one-time costs or anomalies

  7. Model Type: Choose prediction algorithm

  8. Click Update Forecast

Forecast Accuracy

Monitor forecast performance:

  • Mean Absolute Percentage Error (MAPE): Overall accuracy metric
  • Actual vs. Predicted: Historical comparison chart
  • Confidence Score: Model confidence rating
  • Drift Detection: Identify when forecasts become unreliable

Alerts and Notifications

Set up forecast-based alerts:

  • Budget Overrun Alert: Forecast exceeds budget
  • Spike Detection: Unexpected cost increases predicted
  • Accuracy Degradation: Model accuracy drops below threshold
  • Scenario Threshold: Custom scenario-based alerts

Integration with Budgeting

Forecasts automatically integrate with budgets:

  • Proactive Alerts: Warning before budget exhaustion
  • Budget Recommendations: Suggested budget adjustments
  • What-If Analysis: Impact of budget changes on operations

API Access

Access forecast data programmatically:

# Get forecast for account
GET /api/v1/forecast?scope=account&id={account_id}&days=90

# Get forecast accuracy metrics
GET /api/v1/forecast/accuracy

# Generate custom forecast
POST /api/v1/forecast/custom
{
  "start_date": "2026-03-01",
  "days": 60,
  "growth_rate": 0.15,
  "seasonality": true
}

Best Practices

  1. Regular Model Updates: Refresh forecasts weekly or after significant changes
  2. Validate Assumptions: Review growth rates and seasonality patterns
  3. Combine with Budgets: Use forecasts to inform budget planning
  4. Document Scenarios: Keep records of scenario assumptions
  5. Track Accuracy: Monitor forecast accuracy and adjust models accordingly
  6. Account for Changes: Update forecasts after architecture or business changes

Advanced Features

Anomaly Detection

Automatically identify unusual spending patterns:

  • Statistical Outliers: Costs beyond normal variance
  • Trend Breaks: Sudden changes in spending trajectory
  • Service-Level Anomalies: Unexpected service cost increases

What-If Analysis

Model the impact of potential changes:

  • Reserved Instance Purchase: Forecast savings from RI commitment
  • Architecture Changes: Impact of migration or optimization
  • Team Growth: Forecast based on headcount increases
  • New Projects: Add planned new workload costs

Troubleshooting

Forecast Inaccuracy

Issue: Forecasts consistently over or under predict

Solutions: - Increase historical data window - Enable seasonality adjustments - Exclude anomalous historical periods - Consider custom growth rate - Review for recent architectural changes

Missing Forecast Data

Issue: Forecast not available for certain accounts or services

Solutions: - Ensure sufficient historical data (minimum 30 days) - Verify cloud provider connection is active - Check data ingestion status - Confirm account has billable usage