| I&T Solution |
AI L&E Inspection Ranking System
(REF: S-1724) |
| Trial Project |
|
| Solution Feature |
- Data preparation to review with EMSD the candidate data to be incorporated to build the model
- Automated data pipelines will be setup to merge, align, transform and derive data to be used as candidate features
- Model training in supervised/unsupervised settings to formulate a predictive model and highlight significant data features
- Model evaluation to evaluate the accuracy either in back-tests or live-tests.
- Prediction is made regularly (e.g. weekly, monthly) to predict the risk scores of each L&E
|
| Trial Application and Expected Outcome |
- The trial is broken down into several stages: inception (around 2m), fine-tuning (around 3-4m) and implementation (around 2-3m)
- Stage 1: Minimal set of candidate data is used to build the model and to validate the approach with EMSD team
- Stage 2: More candidate data will be added and model parameters will be fine-tuned to enhance the model accuracy/usability
- Stage 3: Automation scripts and pipelines will be built to facilitate the overall data flow
|
| Additional Solution Information |
EMSD-AI-LE-Ranking-V1.pdf
|
| Info on I&T Solution Provider |
|
