Turning a Wasting into a Learning Culture
summary
Claims a reproducible pipeline that turns wasted open text into actionable drivers using real software and numbers: manual coding yields ~70% explanatory power, off‑the‑shelf unsupervised NLP hits only ~36–40%, while a Kalina supervised text-coder matches or exceeds human coding. Causal links come from USM modeled with Bayesian neural networks for causal strength, giving about 40% better explanation than traditional regressions. Output feeds a Predictive Promoter Score dashboard for self‑service what‑if predictions and manager‑friendly key driver charts that identify hidden indirect levers like trustworthiness.
AI-written summary · about
chapters (9)
- 0:00Start
- 5:27Key Driver Analysis
- 6:32The Explanation and Prediction Tool
- 7:28Manual Coding
- 12:49Causal Analysis
- 18:03Universal Structure Modeling Technique
- 18:44Causal Network
- 19:25Bayesian Neural Networks
- 24:10Key Driver Analysis Output
related talks
53mThe Human Side of AI: Unlocking Opportunity in the New Bottlenecks of Software Development
11mTransforming Problem-Solving with Data-Driven Insights
38mFrom code to culture: How AI is reshaping the engineering organization
39mHow to Avoid structural biases in evaluating your Machine Learning/NLP projects
20mGenerative Value Streams: When Code is not a Constraint
21mFine-tune the COE process with generative AI
16mPath to a More Artificially Intelligent Future
31mThe Hidden Costs of Data Quality - Tackling Common Data Challenges in ML