05 · Industry-Sponsored Project
Traxidy Risk Analytics
Turning fragmented project records into cross-project risk signals
Business value
The analysis surfaced resource constraints and scope changes as major delay drivers, helping leadership prioritize interventions.
01 · Business problem
Project performance data lived in 76 SQL tables, and teams defined delay and status differently. Leadership lacked a consistent cross-project view.
02 · My approach
I reduced the data model, standardized GYR and delay definitions, and focused the analysis on actionable root causes.
03 · What I built
Eleven analytical tables, operational KPIs, Random Forest and LSTM models, Tableau dashboards, and a chatbot prototype.
What changed my thinking
The corrections, tradeoffsand decisions behind it.
- 01
More tables did not create more insight
The key move was reducing 76 scattered tables into 11 usable structures.
- 02
Definitions had to be fixed first
A model cannot learn consistently when teams label risk differently.
- 03
The chatbot was an access layer
Its value came from making trusted metrics easier to reach, not from being the main product.