Analytics dashboards

Reports view

The reports view displays long-term summary data rather than individual API events.

Note: The reports view was introduced in v10.0.6, and does not include data that was generated before upgrade to v10.0.6.

Call volume trends provide a high-level view of API activity for a selected time range. The report shows total API calls, successful and failed calls, and the volume of bytes sent and received.

Consumer trends provides an overview of long‑term trends in consumer activity, including application usage, subscription counts, and API call volumes.

AI Platform report analyzes long‑term trends in call volume for AI Platform providers and for consumers of AI LLM and MCP resources.

Consumption report shows long-term trends for API calls processed by gateways, which are essential for consumption-based licensing insights.

Inactive products Identifies products with no subscriptions, no traffic, or no recent traffic

Inactive consumer organizations shows consumer organizations that have no applications, subscriptions, recent traffic, or no traffic at all.

Deprecated products shows products that are deprecated but are still receiving traffic.

The Leaderboard report displays ranked performance data across your selected leaderboard. Select the type of leaderboard you want to view. Click an asset in the table to view more information about it.

Available leaderboard types are:

Products

Ranks all published products by total API call volume.

APIs

Ranks all published APIs by total call volume.

AI Platform providers

Ranks all AI Platform providers by total AI LLM and MCP API calls.

AI Platform consumers

Ranks all AI Platform consumer organizations by total AI LLM and MCP API calls.

Plans

Ranks all published plans by total API call volume.

Applications

Ranks all published applications by total API call volume.

Consumer organizations

Ranks all consumer organizations by total API call volume.

Gateway services

Ranks all gateway services by total API call volume.

Config-sync

Ranks all catalogs by charged value.

Catalog breakdown

Ranks all catalogs in the organization by total API call volume.

Disk usage report tracks and monitors storage usage across APIs, products, applications, and consumer organizations.

Discover view

By default the Reports tab is displayed. To view your API event data, select Discover:

The maximum number of API events that can be accessed from the UI is 10,000. The maximum number of event records that can be displayed per page on the Discover view is 200.

You can customize the columns displayed in the Filters section. You can sort the table in ascending or descending order by clicking on the column headers.

You can switch the timeline chart to a logarithmic y axis, and export the data for the chart by clicking Actions at the upper right of the chart:

You can export the API event data from the table by clicking Actions at the upper right of the table:

If you see unexpected HTTP OPTIONS events in the Discover view, these events are due to CORS requests from browser-based API test tools. For more information about test tools and the OPTIONS call, see Testing an API.

Dashboards view

The analytics dashboards in the API Manager UI show the API event data at the scopes of provider organization, catalog, and space. The scope of the analytics data shown depends on how you access the analytics dashboards, see: Accessing analytics.

You can export each chart in a dashboard as a PNG or JPG image. You can also export the chart’s source data as a JSON or CSV file.

To export a chart, click Actions in the upper-right corner.

It is also possible to enlarge the chart to full-screen by clicking . To see the source data in tabular form click .

Filtering displayed data

You can filter the data that is displayed in the Dashboards and Discover view by defining filters (also called queries). Expand the Filters section to view, define, and apply queries to your displayed data:

Saved queries can be exported as strings for use in REST API and toolkit CLI queries, for example:

copies the following string to the clipboard:

api_name=equals:boo&timeframe=last30days

The Reports interface provides less filters than Discover and Dashboards. In this view, filtering is limited to the time range parameter. Users can also turn the forecast feature on or off. For more information about forecast, see Trend Forecasting

Set a default query in analytics

Trend Forecasting

The Trend Forecasting feature analyzes historical API traffic data to predict future traffic patterns in IBM API Connect Analytics.

Trend forecasting is enabled by default in IBM API Connect v12.1.0.2. You can turn this feature on or off to meet your analysis needs.

How Forecasting Works:

Important: The Trend Forecasting feature uses only mathematical and statistical models. It does not use external machine learning models or AI services. All calculations use established statistical forecasting algorithms.

The model selection process automatically identifies the forecasting model that best matches the characteristics of your historical data. The process evaluates several statistical models and selects the one that provides the best fit for your data patterns.

The Trend Forecasting feature uses the following statistical models for forecasting. Each model is selected based on the characteristics of your historical data.

Table 1. Statistical models

Model	Purpose	Best for	How it works	Example use case
Holt–Winters exponential smoothing	Captures seasonal patterns and trends in time-series data.	Data with recurring seasonal patterns, such as weekly or monthly cycles.	Uses exponential smoothing with three components: Level: The average value of the series Trend: The rate of change over time Seasonality: Recurring patterns at fixed intervals	API traffic that shows consistent weekly patterns, such as higher usage on weekdays and lower usage on weekends.
Day-of-week model	Accounts for weekly cyclical patterns.	Data with strong day-of-week variations.	Calculates average values for each day of the week and projects these values forward.	Business APIs with predictable weekday and weekend usage differences.
Holt’s linear trend smoothing	Captures linear trends without seasonal components.	Data that shows a consistent upward or downward trend.	Uses exponential smoothing with two components: Level: Current value estimate Trend: Rate of increase or decrease	Steady growth in API adoption over time.
Linear regression	Models linear relationships in time-series data.	Data with a clear linear trend.	Fits a straight line through historical data points using the least-squares method. Formula: y = m x + b (m is slope, b is intercept).	Consistent growth or decline in API usage.
Moving average	Smooths short-term fluctuations.	Data with random variations but no strong trend or seasonality.	Calculates the average of recent data points and projects this average forward.	Stable API traffic with minor day-to-day variations.

Understanding Forecasts

Forecasts include confidence intervals that show the range of possible future values:

• Forecast Line: The predicted value (most likely outcome)
• Shaded Area: The confidence interval showing minimum and maximum possible values
• Wider intervals: Indicate greater uncertainty in the prediction
• Narrower intervals: Indicate higher confidence in the prediction

The system provides three confidence levels to help you assess forecast reliability:

• High Confidence: The model has identified strong, consistent patterns in your historical data. Predictions are likely to be accurate. Confidence intervals are relatively narrow.
• Medium Confidence: The model has identified some patterns, but with moderate variability. Predictions provide useful guidance but should be validated. Confidence intervals are moderately wide.
• Low Confidence: Historical data shows high variability or insufficient patterns. Predictions should be used with caution. Confidence intervals are wide, indicating significant uncertainty.
  Note: A notification will appear when forecasts have low confidence.

The following factors influence the confidence level of a forecast:

• Data consistency: More consistent historical patterns lead to higher confidence
• Data volume: More historical data points improve forecast accuracy
• Sudden changes: Traffic spikes or anomalies can reduce confidence
• Seasonality strength: Strong recurring patterns increase confidence

Forecast Horizon and Limitations

Forecast Duration: The forecast duration is determined automatically based on the selected time range.

• Capped at 50% of historical period: Forecasts extend up to half the length of your selected historical time range.
• Maximum limit: The forecast does not extend beyond 30 days, regardless of the length of the historical period.
  Examples:

  • 30-day historical view: Forecast extends 15 days into the future
  • 14-day historical view: Forecast extends 7 days into the future
  • 90-day historical view: Forecast extends 30 days into the future (capped at maximum limit)

Minimum Data Requirements: Forecasting is enabled only when the following data conditions are met:

• Sufficient historical data points: At least several days of data with non-zero values
• Unique dates with data: Multiple distinct dates with recorded activity
• Non-zero values: Forecasting requires actual traffic data (not just zeros)

Insufficient Data Notification: If the selected time range does not contain enough data to generate a reliable forecast, the system displays an informational notification stating that forecasting is unavailable. To resolve this issue, consider the following actions:

• Extend the time range: Include additional historical data to meet the minimum requirements.
• Wait for more data: Allow more activity to accumulate over time.
• Review filters: Verify that applied filters are not limiting the dataset too narrowly.

Enable or disable forecasting

Forecasting is enabled by default starting in IBM API Connect v12.1.0.2. Users can choose whether forecasted data appears in supported dashboards and reports.

Supported dashboards and reports:

• Call Volume Trends
• Consumer Trends
• AI Platform
• Consumer Detail View
• Provider Detail View

Complete the following steps to enable or disable forecasting:

1. Click in the navigation panel.

2. Open a supported dashboard or report.
3. Open the Reports Filter section.
4. Locate the Forecast section, which includes a toggle control.
5. Use the toggle to enable or disable forecasting
   • On (default): Shows historical data plus forecasted values with confidence intervals
   • Off: Shows only historical data without predictions
6. Select the information icon next to the toggle to view additional details.
   Note: The information icon opens a modal with comprehensive details about the statistical models, confidence levels, forecast horizons, and performance considerations.

Interpreting Forecast Visualizations:

• Dashed line: Forecasted values
• Shaded area: Confidence interval (minimum and maximum)
• Today marker: Indicates the current date (boundary between historical and forecast data)

Performance Considerations

Forecasting requires additional computation during chart rendering.

• Large datasets may increase load times.
• Dashboards with multiple forecasted charts may render more slowly.

To optimize performance:

• Disable forecasting when analyzing very large time ranges.
• If charts load slowly, consider the following:
  • Disable forecasting temporarily
  • Reduce the selected time range

Best Practices

To get the most value from trend forecasting:

• Use sufficient historical data: Ensure you have at least 2-4 weeks of historical data for more reliable forecasts
• Consider seasonality: If your API traffic has weekly or monthly patterns, include enough historical data to capture multiple cycles
• Review confidence levels: Pay attention to confidence indicators and use low-confidence forecasts with caution
• Combine with domain knowledge: Use forecasts as one input in your decision-making, not the sole factor
• Monitor accuracy: Compare forecasts with actual results over time to understand their reliability for your specific use case
• Adjust filters appropriately: Overly restrictive filters may result in insufficient data for forecasting
• Plan for capacity: Use forecast upper bounds (confidence interval maximum) for capacity planning to ensure adequate resources
• Disable when not needed: Turn off forecasting when working with very large datasets to improve performance