Treasure AutoML provides multiple solution notebooks: 
Fundamental notebooks include AutoGluon classification/regression, time‑series forecasting, EDA and SHAP explanations, plus notebooks for preparing sample ML datasets. Current solution examples: Next Best Actions, Multi-Touch Attribution, Network Analysis.
The base package includes the fundamental classification/regression notebook using AutoGluon. AutoGluon uses multi‑layer stacking to combine diverse ML models. See the AutoGluon documentation.
It also includes a notebook to prepare sample ML datasets for a quick start.
| Base Package | Description |
|---|---|
| Gluon Train | Train a model using the AutoGluon library on an input training table. |
| Gluon Predict | Use the model created by Gluon Train to predict values. |
| ML Datasets | Load sample ML datasets as TD data tables. |
Pre‑configured solution packages address specific business use cases.
| Solution Package | Description |
|---|---|
| Next Best Action | Predict which marketing action is most likely to increase customer value based on past behavior. |
| Multi-touch Attribution (MTA) | Allocate credit to journey touchpoints toward KPIs (conversion, etc.). |
| Time Series Forecasting | Predict potential future values from historical data to inform strategy. |
| Exploratory Data Analysis | Graphical/statistical analysis to explore data and prepare for modeling. |
| SHAP Analysis | Compute Shapley values for local/global model explainability. |
| Network Analysis | Generate Sankey diagrams and network plots for web access path insights. |
| RFM Analysis | Segment customers by recency, frequency, monetary value (RFM). |
| Clustering | Create clusters (K-means) with feature importance & SHAP explanations. |
| CLTV Prediction | Estimate Customer Lifetime Value (CLTV). |