AgentFarm

Temporal Analysis Module

Module Name: temporal

Analyze temporal patterns, time series trends, periodic behaviors, and temporal efficiency.

Overview

The Temporal module examines how metrics change over time, detecting trends, cycles, and temporal patterns in simulation data.

Key Features

  • Time series analysis
  • Trend detection
  • Seasonality/periodicity detection
  • Autocorrelation analysis
  • Efficiency over time
  • Change point detection

Quick Start

from pathlib import Path
from farm.analysis.service import AnalysisService, AnalysisRequest
from farm.core.services import EnvConfigService

service = AnalysisService(EnvConfigService())

result = service.run(AnalysisRequest(
    module_name="temporal",
    experiment_path=Path("data/experiment"),
    output_path=Path("results/temporal")
))

Data Requirements

Required Columns

  • step (int): Time step/iteration
  • At least one metric column (numeric)

Optional Columns

  • Any metric columns to analyze
  • phase (str): Simulation phase
  • event (str): Special events

Analysis Functions

Analyze temporal trends in metrics.

Outputs:

  • temporal_trends.csv: Trend statistics
  • Trend coefficients, R², significance

analyze_periodicity

Detect periodic patterns and cycles.

Outputs:

  • periodicity_analysis.csv: Cycle detection
  • Period lengths, amplitudes, phases

analyze_autocorrelation

Analyze temporal autocorrelation.

Outputs:

  • autocorrelation.csv: ACF statistics
  • Lag correlations, persistence

analyze_change_points

Detect significant changes in time series.

Outputs:

  • change_points.csv: Detected changes
  • Change locations, magnitudes, directions

Visualization Functions

plot_time_series

Plot metrics over time.

Output: time_series.png

Visualize temporal trends.

Output: temporal_trends.png

plot_periodicity

Plot periodic patterns.

Output: periodicity.png

plot_autocorrelation

Plot ACF/PACF.

Output: autocorrelation.png

Examples

Trend Analysis

result = service.run(AnalysisRequest(
    module_name="temporal",
    experiment_path=Path("data/experiment"),
    output_path=Path("results/trends"),
    analysis_kwargs={
        "analyze_trends": {
            "metrics": ["population", "resources", "efficiency"],
            "method": "linear"
        }
    }
))

Cycle Detection

result = service.run(AnalysisRequest(
    module_name="temporal",
    experiment_path=Path("data/experiment"),
    output_path=Path("results/cycles"),
    analysis_kwargs={
        "analyze_periodicity": {
            "method": "fft",
            "min_period": 10,
            "max_period": 500
        }
    }
))

See Also

Module Version: 2.0.0