Guides

Times of Interest (TOIs)

Temporal parameters that define when your analysis should run

Times of Interest (TOIs)

Times of Interest (TOIs) define the temporal parameters for your analysis - when your analysis should run and what time periods to analyze. They answer the "When?" question of the three fundamental questions framework.

What are TOIs?

TOIs are time ranges and temporal cadence specifications that determine:

  • Start and end times for analysis
  • Temporal grouping (hourly, daily, weekly, etc.)
  • Recurrence patterns
  • Temporal filtering of input data

TOI Components

Time Range

Defines the overall period for analysis:

  • Start time: When the analysis period begins
  • End time: When the analysis period ends
  • Duration: Can span from hours to years

Temporal Cadence

Defines how to group time within the range:

  • Hourly: Analyze each hour separately
  • Daily: Group by day
  • Weekly: Group by week
  • Monthly: Group by month
  • Custom: Define specific time intervals

Recurrence Rules

Optional patterns for selective time periods:

  • Weekdays only: Monday through Friday
  • Business hours: e.g., 9am-5pm
  • Specific days: e.g., weekends, holidays
  • Custom patterns: Any recurring schedule

Use Cases

Real-Time Monitoring

TOI: Last 24 hours with hourly cadence

  • Monitor current conditions
  • Track recent changes
  • Enable rapid response

Historical Analysis

TOI: Full year with daily cadence

  • Identify seasonal patterns
  • Track long-term trends
  • Perform year-over-year comparisons

Business Hours Analysis

TOI: 30 days, weekdays 9am-5pm only

  • Focus on operational periods
  • Exclude off-hours noise
  • Optimize processing costs

Trend Detection

TOI: 5 years with monthly cadence

  • Identify long-term patterns
  • Track gradual changes
  • Understand seasonal cycles

Temporal Filtering

TOIs enable temporal filtering of input data:

1. Platform identifies all available data
         ↓
2. Filters data to only include timestamps within TOI range
         ↓
3. Groups data according to temporal cadence
         ↓
4. Passes each time group to algorithm
         ↓
5. Algorithm processes data for each time period

This ensures:

  • Focused analysis: Only relevant time periods processed
  • Organized results: Data grouped by meaningful time intervals
  • Efficient processing: No wasted computation on excluded times

Parallelization by Time

Time periods can be processed independently, enabling:

  • Parallel execution: Multiple time periods processed simultaneously
  • Faster results: Reduces total processing time
  • Scalable architecture: Distribute workload across containers

Example: Analyzing hourly data for a full day (24 hours) could run 24 parallel computations if all data is available simultaneously.

Combined Parallelization

TOIs work with AOIs for two-dimensional parallelization:

3 AOIs × 24 hours = 72 parallel processing units

This enables:

  • Maximum throughput: Process multiple locations and times concurrently
  • Flexible scaling: Adapt resources to workload
  • Cost efficiency: Complete large analyses quickly

TOI Configuration Examples

Example 1: Recent Activity

Start: 7 days ago
End: Now
Cadence: Daily
Result: 7 daily snapshots of recent activity

Example 2: Business Quarter Analysis

Start: January 1, 2024
End: March 31, 2024
Cadence: Weekly
Recurrence: Weekdays only
Result: ~13 weekly summaries, excluding weekends

Example 3: Hourly Monitoring

Start: Today at midnight
End: Today at 11:59pm
Cadence: Hourly
Result: 24 hourly analyses for the current day

Example 4: Multi-Year Trend

Start: January 1, 2020
End: December 31, 2024
Cadence: Monthly
Result: 60 monthly data points over 5 years

Best Practices

Cadence Selection

  • Match to data availability: Don't use hourly cadence if data updates daily
  • Balance detail vs. cost: Finer granularity = more processing
  • Consider analysis goals: Use appropriate time scale for insights

Time Range Considerations

  • Data availability: Ensure data exists for the specified range
  • Processing time: Longer ranges with fine granularity take more time
  • Cost implications: More time periods = more processing units

Recurrence Patterns

  • Business relevance: Focus on operationally meaningful times
  • Cost optimization: Exclude irrelevant time periods
  • Comparative consistency: Use same patterns for fair comparisons

Creating TOIs

TOIs can be created through:

Elements Application

  • Select time ranges using date pickers
  • Choose cadence from predefined options
  • Configure recurrence rules visually

Elements API

  • Define programmatically using the SDK
  • Specify ISO 8601 timestamps
  • Use flexible recurrence rule syntax

Temporal Data Types

Different data sources have different temporal characteristics:

Satellite Imagery

  • Revisit time: Days to weeks between captures
  • Best cadence: Daily or weekly
  • Considerations: Cloud cover, sensor availability

Device Location Data

  • Frequency: Continuous or near real-time
  • Best cadence: Hourly to daily
  • Considerations: Data density, privacy windows

Aerial Imagery

  • Frequency: On-demand or scheduled flights
  • Best cadence: As available
  • Considerations: Flight schedules, weather

Next Steps

  • Define geographic regions with AOIs
  • Learn about Algorithms that process temporal data
  • Combine into workflows with Analyses
  • Execute on specific time periods through Computations

Updated 5 months ago