Figure 1: Dependence relationship among study variables.
The figure shows a directed acyclic graph with the following relationships: epsilon -> y upsilon -> w w -> y x -> w x -> y
Figure 2: Root mean squared error feature contribution for house price
control variables.
The figure shows a horizontal bar chart with control variables listed on the vertical axis and change in root mean squared error on the horizontal axis. The most important feature is square footage, followed by sale date and distance to the coast.
Figure 3: House price variations by time and location
(a) Partial dependency plot of log house price on sale
date.
(b) Relative predicted price of a house with median
characteristics by location.
The figure contains two panels. On the left panel is an upward sloped line showing how log house price varies by sale date. On the right panel is a map of Plorida with house price percentiles colored from lowest (green) to highest (red). The highest price markets are near Miami and the Florida keys.
Figure 4: Partial dependency plot of log house price on structure
size.
The figure shows a sharply upward sloped curve at levels off at the top. The figure indicates that log house price increases rapidly from 500 to 5,000 square feet but little after that point.
Figure 5: Partial dependency plots of log house price on selected
house features.
The figure shows 9 panels of bar charts showing how different categorical variables predict house prices.
Figure 6: Relationship between non-flood weather average annual loss
and effective insurance premium rate.
The figure shows a scatter plot with an upward sloped regression line indicating that log AAL is highly positively correlated with log insurance premium.