Back to Solution Workbench

Agentic Data Science & Insights Suite

Autonomous 6-phase analytical orchestration: from business hypothesis framing and encrypted edge profiling to gradient boosted inference and human-in-the-loop governance sign-off.

Interactive Live Scoring Engine

In-Memory Sub-15ms Inferencedonor_survival.js (Weibull Hazard)

Real-time statistical evaluation executing live in-browser across production-scale cohorts. Adjust the 3 model threshold sliders below to observe instantaneous recalculations of at-risk capital, flagged entity cohorts, and recoverable ROI.

At-Risk Capital$15,107
Flagged Entities16 Donors (64%)
Recovery Potential$8K–$11K
Engine Latency0ms
Inactivity Threshold120 days
45d (Aggressive)120d (Baseline)240d (Lenient)
Email Engagement Weight45%
5% (Low Dampening)45% (Default)100% (High Credit)
Event Retention Hedge1.25×
0.50× (Mild Hedge)1.25× (Calibrated)2.50× (Max Hedge)
Select Production Analytics Scenario
Zero Client Data RetentionReady for Execution
Control Panel:
Current Phase:01 — Decide

Phase 01

Decide

Live

02

Phase 02

Connect

03

Phase 03

Analyze

04

Phase 04

Interpret

05

Phase 05

Recommend

06

Phase 06

Review

P-01Processing

Decide

Hypothesis & metric contract

P-02Pending

Connect

Encrypted source profiling

P-03Pending

Analyze

Gradient boosting & SHAP

P-04Pending

Interpret

Driver attribution narrative

P-05Pending

Recommend

Policy playbooks & ROI

P-06Pending

Review

HITL governance sign-off

Business Objective Framing

Phase 01 — Hypothesis Definition & Target Contract

Contract Verified

Identify key drivers of donor churn for a regional community foundation to proactively intervene and protect at-risk annual giving revenue across the donor portfolio.

Target Metric

Donor churn (12-month binary classification)

Ground-truth validation criteria

Observation Window

2025-01-01 → 2025-12-31

Granularity: Monthly

Explanatory Candidates

8

High-cardinality candidate signals

Tested Empirical Hypotheses

H1:Inactivity exceeding 120 days exponentially increases churn odds across all giving tiers.
H2:High-value donors churn due to lack of personalized touchpoints rather than giving capacity.
H3:Event attendance serves as the strongest behavioral retention hedge.

Bounded Explanatory Feature Vector

days_since_last_donationtotal_donation_amount_lifetimeavg_donation_frequency_monthlyemail_open_rate_90devent_attendance_countdonor_tenure_monthsgift_type_recurringgeographic_region