AlarmAnalysis/CLAUDE.md

Alarm Analysis

Python project for analyzing alarm data from building monitoring systems (CSV alarm logs + XLSX sensor reports).

Commands

# Activate virtual environment
source alarm_analysis_env/Scripts/activate  # Windows Git Bash
# or: alarm_analysis_env\Scripts\activate   # Windows CMD

# Run full analysis (no plots)
python run_analysis.py

# Generate plots (requires matplotlib display)
python create_plots.py

# Run tests
python test_changes.py
python test_duration_fix.py
python test_mapping.py
python test_enhanced_plotting.py

Architecture

• alarm_analyzer.py — Main AlarmAnalyzer class (~96KB). Handles data loading, alarm categorization, event pairing, duration calculation, basic/advanced analysis, uptime metrics, and export.
• run_analysis.py — Entry point that runs the full pipeline without visualizations.
• create_plots.py — Generates alarm dashboard, duration analysis, and sensor analysis plots.

Data Files

• CardinalAlarmsDec25.csv — Raw alarm data (columns: Alarm_Id, Sensor_Id, Date, Description, LogTime)
• SensorReport Cardinal 2025-12-23_processed.xlsx — Sensor descriptions and group mappings
• exclusion_config.json — JSON format: {"excluded_groups": ["GroupName1"]}
• groups_to_skip.txt — Text format: one group name per line

Key Patterns

• Alarm types are parsed from Description field via regex: Hi/Lo Alarm, Hi/Lo Warning, Error, Normal
• Events are paired (alarm start -> Normal end) to calculate durations
• Sensor mapping links Sensor_Id to human-readable names and groups from the XLSX file
• Visualization imports are deferred (_import_viz_libs()) so analysis can run headless
• Output goes to output/ (CSVs) and plots/ (PNGs)

Dependencies

Python 3.13 with: pandas, numpy, matplotlib, seaborn, openpyxl