Fortune 500 Risk Simulation based on PGE Bankruptcy Data for Wildfire RIsk

Save to Favorites

Recruiting Organization: M4A FOUNDATION - CROWDDOING

Fortune 500 Risk Simulation based on PGE Bankruptcy Data for Wildfire Risk

Prevention Derivatives draws upon the annual recurring expected losses of Fortune 1000 companies. It would be helpful to do more with the data on the PGE bankrupsy to fully simulate the 75 Fortune 500 companies that expect that loss to recur at 2.5% to 10%.

-Prevention derivatives is driven by the thesis that there is an under-valuation of passive risk (or the cost of inaction) and an under-prioritization of positive risk. Correspondingly for wildfires as an example, there is an under-recognition of the potential shared value upside of preventative action through social innovation and social interventions (such as goats & sheep that prevent wildfires). CrowdDoing.world's aim is to guarantee positive risk through leveraging existing liabilities to allow for the implications of prescriptive analytics to be financed. The under-pricing of passive risk means that liabilities are treated as either costs of doing business or un-predictable risks even for entirely preventable risks. Risk management offices have been too biased towards avoiding taking the wrong risks rather than ensuring that institutions make their own luck by seizing the abundant positive risk opportunities in social innovation. Meanwhile, the bias against positive risk leaves social innovations not to get adopted even if there would be remarkable benefits to all stakeholders if they were adopted

In the framework of Prevention Derivatives, we want to create a predictive machine learning (ML) model that for a given geographical region will estimate likely savings (losses) due-to protection (damages) of stakeholders’ properties, business profits, common health, and regional ecology resulting in applying risk prevention solutions (or doing nothing instead). Goal of these notes is to analyze ML model’s design, offer a potential improvement and to discuss existing approaches for data collection, and training and testing the model. It is important to notice that the model is applied to the entire selected or target region. Therefore, a geographical region R is the smallest unit we apply modeling to.

Data science will be utilized in the following ways:

Explore/Visualize data currently available on Wildfires

Identify trends and patterns in Historical data

Quantify historical losses in dollars based on property destruction, casualties, acres burnt, etc.

Build predictive models to identify areas of high wildfire risk based on factors such as weather, vegetation, topography, etc.

Visualization of Model outcomes

Scenario building (changing input variables and observing impact on outcome)

Tools - R, Python, MATLAB, SQL, PowerPoint

Knowledge or Interest in anyone or more:

Identify papers on Simulation of Wildfires, Catastrophe Modeling

Review and present Technical papers in a way that everyone can understand

Assist in Model development and testing by contributing in finding data and programming

Identify/Collect data relevant to wildfire Impact

Work cross-functionally

Traits:

Mathematically inclined, highly analytical, creative problem solver, can conduct analyses independently or with minimal supervision

Programming for Data Science

Mathematics

Statistics

Predictive Analytics

Prescriptive Analytics

Machine Learning - Supervised/Unsupervised learning

Artificial Intelligence

Data Mining

Computer Science

Monte Carlo Simulations

Expectations:

If you have any questions about processes for joining CrowdDoing.world as a volunteer to support our efforts in systemic change please write to volunteerorientation@crowddoing.world

Stipend ProvidedFalse

Training ProvidedFalse

Housing AvailableFalse

Language/Cultral Support AvailableFalse

Wheelchair AccessibleFalse

Fee RequiredFalse

Fee Amount: None

This opportunity is recurring