Analytics
Journal Entry Risk Analytics
Work in progress
Nov 2025 – Present
Auditor-facing pipeline that scores and surfaces risky general ledger journal entries, pairing anomaly detection with explainable risk factors (full report coming soon).
- Engineers journal-entry features (amount spikes, period-end timing, user frequency, approvals) and standardizes them into a model-ready dataset.
- Runs scikit-learn anomaly detection (Isolation Forest + Local Outlier Factor) to flag high-risk postings and label them with interpretable drivers.
- Exports risk tiers and supporting evidence to CSV/SQLite so auditors can triage entries quickly and trace decisions.
Analytics
Spotify Data Analysis
Completed
Nov 2025 – Dec 2025
Exploratory analysis of Spotify listening history to surface top artists, session patterns, and audio-feature trends, pairing export data with API lookups and reproducible notebooks.
- Aggregated raw Spotify streaming exports and enriched tracks with API audio features (energy, danceability, tempo) to create a unified plays dataset.
- Profiled listening habits by hour, weekday, and device to flag high-engagement windows and shifts in genre/artist mix over time.
- Built visual reports (heatmaps, feature distributions, cumulative play curves) and documented insights in notebooks for repeatable reruns.
Dashboard
Telecom KPI Command Center
Completed
Dec 2025
An executive-ready dashboard for telecom ops leaders that surfaced network health, ticket velocity, and outage hotspots.
- Modeled KPIs with DAX and built drill-through views for regional and device-level insights.
- Automated refresh pipelines so reports stayed current without manual exports.
- Packaged findings into a one-page briefing that accelerated weekly decision meetings.
Analytics
VCT 2024 Analysis
Completed
Dec 2024 – Jan 2025
Valorant event analysis pipeline that pulls raw match data from VLR, shapes it into agent- and match-level CSVs, and surfaces composition and regional insights (full report coming soon).
- Pulls VLR events via vlrdevapi (e.g., Champions 2024) and turns them into agent- and match-level datasets so raw pages become structured rows you can analyze.
- Builds per-team composition summaries for every match and map, then highlights the most common comps to spot meta trends quickly.
- Rolls up region-by-region stats showing which agents and compositions get picked and how often they win, making it easy to compare regions at a glance.