CrisisLens.ipynb

michaelml.dev / Michael_Portfolio.ipynb

🏆 1st Place Winner at Hacklytics (Databricks x United Nations Challenge). CrisisLens is an AI-assisted crisis intelligence platform that combines real-time geospatial context, ML forecasting, and Databricks-powered natural language analysis to help humanitarian teams prioritize aid decisions.

[ ]

code · CrisisLens

project = load_project("crisis-lens")
project.summary()

Output

CrisisLens

DatabricksUnited Nations ChallengeNext.jsTypeScriptThree.jsTailwind CSSPyTorchLightGBMXGBoostPandasNumPyForecasting

[ ]

code · Problem definition

project = load_project("crisis-lens")
project.problem()

Output

ProblemHumanitarian teams must prioritize aid while crisis, funding, and operational signals are fragmented across countries and time.

ApproachFuse multi-country data with leakage-aware ensemble forecasting, SHAP analysis, KNN peer mapping, Databricks Genie, and an interactive 3D globe.

OutcomeWon 1st Place in the Databricks × United Nations Challenge at Hacklytics and delivered a working crisis-prioritization command center.

[ ]

code · Constraints

project = load_project("crisis-lens")
project.constraints()

Output

Constraints

Heterogeneous country-level data needed a consistent schema.
Time and group leakage had to be prevented during validation.
Forecasts needed to remain understandable inside an interactive decision tool.

[ ]

code · Role and implementation

project = load_project("crisis-lens")
project.implementation()

Output

Role and implementation

My role: Cleaned and structured the datasets; performed EDA, feature engineering, and SHAP analysis; trained and validated the forecasting ensemble; built peer mapping; and integrated predictions into the visualization.

Built 6-month and 12-month-plus forecasting workflows with multiple ensemble models.
Used SHAP for feature analysis and KNN for cross-country peer mapping.
Connected model artifacts, Databricks analysis, simulations, and a Three.js interface.

[ ]

code · Architecture

project = load_project("crisis-lens")
project.show_architecture()

Output

Data FusionAggregate and normalize multi-country humanitarian indicators into dashboard-ready artifacts.

ModelingTrain ensemble and multi-horizon forecasting models with leakage-safe, time-aware validation.

InterpretabilityUse SHAP and comparative analysis to surface drivers of country-level risk and funding shifts.

Genie Analysis ModeEnable natural-language Databricks queries that return structured summaries and comparison tables.

Simulation ModeRun intervention scenarios and visualize projected quarter-by-quarter impact on a live globe.

Operational UIPresent model outputs through an interactive command center with geospatial overlays and controls.

[ ]

code · Demo and media

project = load_project("crisis-lens")
project.media()

Output

[ ]

code · Evaluation

project = load_project("crisis-lens")
project.evaluation()

Output

Evaluation

Applied group- and time-aware cross-validation.
Compared LightGBM, Random Forest, XGBoost, Gradient Boosting, and stacking approaches.
Reviewed feature drivers with SHAP rather than treating predictions as opaque scores.

[ ]

code · Results

project = load_project("crisis-lens")
project.results()

Output

Results

Produced multi-horizon country forecasts and scenario outputs inside the live interface.
Awarded 1st Place in the Databricks × United Nations Challenge at Hacklytics.

[ ]

code · Tradeoffs and failures

project = load_project("crisis-lens")
project.tradeoffs()

Output

Tradeoffs

The ensemble improves coverage at the cost of model and deployment complexity.
Scenario outputs support decisions; they are not a substitute for field intelligence.

[ ]

code · What I would improve

project = load_project("crisis-lens")
project.improvements()

Output

Improvements

Add calibrated uncertainty intervals and more explicit data-quality indicators.
Validate prioritization workflows with humanitarian-domain users.

[ ]

code · Links and repository

project = load_project("crisis-lens")
project.links()

Output

GitHub Demo Project page

CrisisLens

DatabricksUnited Nations ChallengeNext.jsTypeScriptThree.jsTailwind CSSPyTorchLightGBMXGBoostPandasNumPyForecasting

Role and implementation

Built 6-month and 12-month-plus forecasting workflows with multiple ensemble models.

Used SHAP for feature analysis and KNN for cross-country peer mapping.

Connected model artifacts, Databricks analysis, simulations, and a Three.js interface.