Embedded Files
Data Science project
for client Zurich Insurance
Zurich Insurance Plc
Client: Zurich Insurance
Zurich is a leading multi-line insurer serving people and businesses in more than 200 countries and territories and has about 60,000 employees. Founded more than 150 years ago, Zurich is transforming insurance.
Helped grow the Data Science department and capabilities in Zurich Insurance Spain.
Worked directly under the Head of Data and in close relationship with the CEO and the other heads of departments depending on the topics and projects handled.
Was among the promising youth in the company involved in strategic and innovative company projects
ZAgents
Project description
Led the project ZAgents (Zurich Agent Dashboard) which is a dynamic, data-driven platform designed to provide a comprehensive visualization of insurance agents across Spain.
The dashboard integrates multiple data sources to offer in-depth agent profiles, perform distance calculations, and apply clustering and classification techniques to identify potential mergers and acquisitions opportunities.
The solution received highly positive feedback from the business, leading to iterative refinements to meet evolving expectations.
Leveraging Google Maps for its intuitive interface and user-friendly design, we delivered a powerful tool that seamlessly presents actionable insights and supports strategic decision-making.
Role & Responsibility
Lead Data Scientist responsible for overseeing the end-to-end data science project lifecycle, including data collection, defining problem statements, and developing robust predictive models. Key responsibilities include:
- Collaborating with cross-functional teams to identify business challenges and translate them into actionable data-driven solutions.
- Designing and implementing data collection strategies to gather relevant and high-quality datasets.
- Building, validating, and deploying machine learning models to address specific business needs.
- Conducting exploratory data analysis to uncover insights and inform decision-making.
- Establishing and maintaining strong relationships with stakeholders, ensuring alignment between data initiatives and organizational goals.
- Mentoring and guiding junior data scientists in best practices, methodologies, and analytical techniques.
- Communicating complex findings and recommendations to non-technical stakeholders in a clear and impactful manner.
- Continuously monitoring model performance and iterating on solutions to optimize outcomes.
Accomplishment
Developed an interactive map using Google Maps to visualize all agents and brokers, integrating ownership details, financial data, and geographic location for comprehensive analysis.
Leveraged geocoding technologies such as HERE.com and Google Places Geolocation API to enhance data accuracy and usability.
The dashboard identified 25 potential mergers and 15 acquisition opportunities, which were further investigated by the management team, significantly contributing to strategic business decisions.
Google Maps representation of the final version and with all details computed
Explanation of libraries
Pandas: For data manipulation and analysis, especially to handle the agent, broker, and financial data.
Geopandas: For working with geospatial data and performing geographic operations.
NumPy: For efficient numerical operations and array manipulation.
Matplotlib / Plotly: For visualization, especially if you want to visualize data trends or plot additional insights on the map.
Folium: A wrapper around the Leaflet.js library, used to visualize geospatial data on an interactive map (can be used in addition to Google Maps for flexibility).
gmplot: A simple wrapper around Google Maps to plot geospatial data easily.
Google Maps API Client (googlemaps): To interact with Google Maps services for tasks like geocoding and reverse geocoding.
requests: For making API calls to services like Google Maps and HERE.com.
Scikit-learn: For clustering, classification, and other machine learning tasks like identifying patterns or segmenting the agents/brokers.
H3: Uber's open-source spatial indexing library for dividing the earth into hexagonal grids, useful for geographic clustering.
Shapely: For geometric operations on spatial data, especially when working with geographic boundaries.
Explanation:
Clustering Algorithms:
K-Means Clustering: For grouping agents or brokers based on geographic proximity or financial data.
DBSCAN (Density-Based Spatial Clustering of Applications with Noise): For identifying spatial clusters with arbitrary shapes (useful for geographic data).
Agglomerative Hierarchical Clustering: For identifying clusters of brokers and agents by merging the nearest clusters based on their geographic and financial data.
Geospatial Algorithms:
Haversine Formula: To calculate the great-circle distance between two points on the Earth's surface, used for distance calculations between agents and brokers.
Geocoding Algorithms: To convert addresses into geographic coordinates, using Google Maps Geocoding API or HERE.com.
Classification Algorithms:
Random Forest / Decision Trees: For classifying agents and brokers based on their financial performance, location, and other relevant data.
Logistic Regression: If you need a binary classifier to identify potential merger/acquisition candidates.
Google Sites
Report abuse