Incubated concepts for human-in-the-loop pipelines and data driven segmentation strategies to increase contact data quality to 90% accuracy (from 80%). Primary driver for NPS growth (+10) in 18 months (2017). Key modules include:
- Employment churn predictors models using LightGBM on features like: company size, industry, distinct companies, employment history, function, level, etc...)
- Confidence score ranking models on features like: update recency, source, richness, multi-source, etc...
- Advanced segmentation and automatic workflows
- Multi-threaded sources using:
- Crowd annotations
- Open web harvests
- Corp domain text mining
