incident-analysis-nci 1.0.0

Neighbourhood Safety Incident Reporting analysis library

incident-analysis-nci

A Python library for classifying, validating, analysing, and formatting neighbourhood safety incident reports.

Author: Adithya Reddy Version: 1.0.0

Installation

pip install incident-analysis-nci

Or install from source:

pip install -e .

Features

• IncidentClassifier — Classify incidents by type, severity, and priority using keyword matching
• ReportValidator — Validate and sanitise incident reports and user data
• SafetyAnalytics — Generate statistics, hotspot maps, trend data, and safety reports
• IncidentFormatter — Format incidents for display, CSV export, and SNS alert notifications

Quick Start

from incident_lib import IncidentClassifier, ReportValidator, SafetyAnalytics, IncidentFormatter

# Classify an incident
clf = IncidentClassifier()
category = clf.classify_incident("Someone stole my bike from the rack")
print(category)  # "theft"
print(clf.get_severity_level(category))  # "high"

# Validate a report
val = ReportValidator()
valid, errors = val.validate_report({
    "title": "Bike stolen on Main St",
    "description": "My mountain bike was stolen from outside the shop at 10pm.",
    "category": "theft",
    "location": "Main Street",
})
print(valid)  # True

# Analytics
analytics = SafetyAnalytics()
stats = analytics.get_incident_stats(incidents)
hotspots = analytics.get_hotspot_areas(incidents)
report = analytics.generate_safety_report(incidents, "Dublin 1")

# Formatting
fmt = IncidentFormatter()
print(fmt.format_incident_summary(incident))
print(fmt.to_csv(incidents))
print(fmt.format_alert(incident))

Dependencies

None — uses Python standard library only (re, datetime, csv, io, collections, math).

Running Tests

python -m pytest tests/ -v

Or with unittest:

python -m unittest discover tests -v

Incident Categories

Category	Severity
assault	critical
fire	critical
theft	high
accident	high
flooding	high
vandalism	medium
suspicious_activity	medium
noise_complaint	low
other	low

License

MIT