Project overview

• Develop a model and an interactive pricing tool for an agricultural insurance product using open agro-climatic datasets and remote sensing data.
• Focus on combining agro-climate variables (precipitation, temperature, soil, vegetation indices) with historical loss/exposure proxies to produce risk scores and premium pricing.

Objectives and tasks

• Design and implement statistical and machine learning models (e.g., regression, random forest, time-series models) to estimate yield/risk and to derive insurance premium curves.
• Build an interactive pricing simulation (Streamlit or Dash) that allows stakeholders to explore pricing under different scenarios and input parameters.

Technical environment and data sources

• Languages: Python or R; recommended libraries include scikit-learn or statsmodels for modeling and pandas/xarray for data handling.
• Data sources: WorldClim (climate layers), SoilGrids (soil properties), FAO (agricultural statistics), CHIRPS (precipitation), Sentinel (remote sensing indices). Use PostgreSQL for structured data storage and Git for version control.

Implementation details and optional infrastructure

• Use Streamlit or Dash to create an interactive pricing simulator; integrate model outputs and visualizations (maps, time series, uncertainty bands).
• Optionally use Google Colab or AWS EC2 for scalable computation and heavier preprocessing of Sentinel imagery.

Expected deliverables

• A reproducible modeling pipeline (code, notebooks or scripts) that ingests open data, preprocesses agro-climate variables, trains models, and outputs risk estimates.
• An interactive pricing application (Streamlit/Dash) plus documentation explaining assumptions, model validation results, and recommended pricing strategy.

Required skills and desirable experience

• Strong proficiency in Python or R and experience with machine learning/statistical modeling (scikit-learn, statsmodels, or equivalent).
• Experience handling geospatial and time-series agro-climatic data (CHIRPS, WorldClim, Sentinel) and familiarity with PostgreSQL.

How to apply

• To apply for this specific project, send your CV and a short motivation note to stages@hydatis.fr referencing "Clim-03 Modeling and Pricing an Agricultural Insurance Product" in the subject line.
• You may also visit the company website for more information: https://www.hydatis.com