pandas_skill

Overview

This skill provides expert pandas-based data analysis and manipulation tailored for customer support operations. It delivers reusable patterns for SLA tracking, ticket trend analysis, agent performance metrics, ETL integration with PostgreSQL, and production-ready data cleaning. Use it to turn raw support logs into actionable dashboards and automated reports.

How this skill works

The skill inspects support ticket datasets and performs DataFrame creation, type optimization, indexing, and time series resampling. It applies cleaning and validation, computes SLA and agent metrics, runs groupby aggregations (including custom percentiles), and supports loading/saving via SQLAlchemy to PostgreSQL. Outputs include cleaned datasets, aggregated metrics, compliance tables, and validation reports.

When to use it

• Track SLA compliance and generate SLA dashboards
• Analyze ticket volume trends and seasonality
• Measure and compare agent and team performance
• Integrate ticket data from PostgreSQL into analytics pipelines
• Automate cleaning, validation, and metric exports to a database

Best practices

• Parse and set datetime columns early; use datetime index for time-series operations
• Cast low-cardinality text to category to save memory and speed grouping
• Use vectorized operations and .assign() for readable, chainable transformations
• Validate and log data quality: missing fields, duplicates, and invalid timestamps
• Use resample + rolling windows for trend smoothing and pct_change for momentum

Example use cases

• Calculate response and resolution SLA compliance by priority and produce compliance rate tables
• Resample tickets by day/week/month to detect spikes, compute 7/30-day rolling averages, and percent changes
• Compute per-agent metrics: tickets handled, avg/median/95th response times, CSAT, reopen rates, and tickets/day
• Clean raw ticket exports: fill unassigned agents, standardize priorities, drop invalid rows, detect outliers and produce a validation report
• Load filtered ticket ranges from PostgreSQL, process metrics, and write summarized metrics back to a metrics table