Risk Analysis by the Numbers Atlantic Tropical Cyclones Jonathon Carrasco University of Houston Bauer College of Business Department of Finance
Questions to Address How to merge data from various sources for a more complete picture of “what is going on” How to make a little data say a lot – Graphs – Statistics How to interpret all of the above
Sources Willis Energy Loss Number of Incidence Actual Losses US$ Indexed Losses US$ The National Hurricane Center's Tropical Cyclone Number of Tropical Storms Tropical Storm Classification Tropical Storm Predicted Named Storms Predicted Hurricanes Predicted Intense Hurricanes
Construction of Box Plot Middle Line - Median Value The box is made up of the 25 th and 75 th percentile. This is also referred to as the “inner quartile range” Data points that are more than 1.5 times away from the “inner quartile range” are considered outliers. “Whiskers” are assigned to the lowest non-outlier Dots are assigned to values that lie beyond the whiskers. Open dots indicate “mild outliers” (1.5< X< 3). Closed dots indicate “extreme outliers” (3 < X) Flickr: “Boxplot” (waynew47)
Visualize as a Bell Curve
Hurricanes
Tropical Storms
Storm Classification Hurricane Tropical Storm Tropical Depression Subtropical Storm Subtropical Depression
Take a look at the Data Year Subtropical Depression Subtropical Storm Tropical Depression Tropical Storm Hurricane
Create an Intensity Score Year Subtropical Depression Subtropical Storm Tropical Depression Tropical Storm Hurricane Intensity Score Storm ClassificationIntensity Contribution Subtropical Depression1 Subtropical Storm2 Tropical Depression3 Tropical Storm4 Hurricane5
Intensity Score
Total Losses
Total Losses (Log10)
Regression Analysis Drawing a line through a scatterplot Minimizes the least- squares error y = a + Bx + e
Statistical Tests t-test – How far away from the mean divided by the estimated variance (noise) in the sample F-test
Breakdown of a regression
Intensity Analysis
Predicted Tropical Storms