Exposure Data
Data Quality
Geocoding
Hazard Retrieval
Grab Bag
100
Exposure data is data that is used to describe: - WHAT is re/insured - HOW it is re/insured - WHERE it is located - AND ??????
What is how much it's worth/value
100
If you are looking at a policy for a retail grocery store chain and you do not have construction information, without degrading the quality of the data, you should: A) Make a conservative assumption about the construction type in case the buildings are more vulnerable than the average. B) Leave the construction type as unknown, unless you know something about how stores in this chain are built (in which case you could set a construction type default). C) Try to find a picture of a typical grocery store in this chain and make an assumption about the construction type. D) None of the above
What is B) Leave the construction type as unknown, unless you know something about how this chain of retail grocery stores is built (in which case you could set a construction type default).
100
Geocoding is fundamentally defined as: A) Cleaning and standardizing exposure data. B) Certifying that location data follows local postal service formats. C) Estimating the latitude and longitude of a location based on its street address, city, or postal code. D) None of the above responses are accurate definitions.
What is C) Estimating the latitude and longitude of a location based on its street address, city, or postal code.
100
True or False: Hazard retrieval is performed after damages are calculated in a catastrophe model.
What is False. Hazard data values are one of the inputs into the damage calculation portion of the model, so hazard retrieval must happen before damages are calculated.
100
Overall, RMS hurricane frequency rates are higher or lower than the historical average
What is higher
200
Which of the following are inappropriate uses of data profiles? A) Cost allocations B) Expanding into new territories C) Ratemaking D) Exposure data profiles may be used for any of the above
What is D) Exposure data profiles may be used for any of the above
200
True or False: Obtaining the highest possible exposure data quality, while desirable, is not always necessary given the type of catastrophe risk being analyzed.
What is True. Data quality should be defined around its fitness for purpose. This sometimes means that a lower data quality for a specific parameter is sufficient for its intended purpose (e.g. a geocoding level of postal code may be sufficient for a particular peril/region catastrophe risk analysis)
200
As geocoding resolution improves, modeling uncertainty: A) Increases B) Decreases C) Stays the same
B) Decreases. If the geocoder performs a match at a high resolution, such as the street or building level, there is relatively high confidence in the precision of the latitude and longitude, and there will be more certainty in the accuracy of the loss results.
200
The primary role of geocoding and hazard retrieval in catastrophe modeling is: A) To create a map of exposure distribution. B) To link address data with the geography of a model. C) To guarantee that a location will generate losses. D) To assign latitude and longitude values.
What is B) To link the address data with the geography of a model.
200
The historical hurricane average frequency rates are higher or lower than the Brit view of frequency rates
What is lower
300
True or False: If I am a Specialty Insurance carrier, it is always best, when dealing with unknown data, to utilize industry inventory databases for weighted averages of the data.
False. Since I fill a niche and do not necessarily follow the industry-wide averages, there may be instances where assumptions will improve the quality of the data.
300
Assigning more than six secondary modifiers to every location in a portfolio: A) Always increases the quality of your catastrophe analysis results B) May be a data quality flag requiring further exploration of either data bias or data completeness C) Requires a field check to assure data accuracy, but assigning six or less does not D) All of the above E) A and B only
What is B) Assigning more than six secondary modifiers to every location in a portfolio may be a data quality flag requiring further exploration of either data bias or data completeness.
300
Software algorithms for address standardization can assist with which of the following common errors in location data? A) An analyst incorrectly enters “Gotenba” into the model due to a transliteration error instead of the correct name “Gotemba”. B) An analyst incorrectly enters “12 3mains T” into the model due to a data parsing error instead of the correct address “123 Main St.” C) An analyst incorrectly enters “123 Mian St” into the model due to a typographical error instead of the correct address “123 Main St.” D) All of the above would be likely be fixed by geocoding software algorithms. E) A and C are both correct.
E) A and C are both correct. If due to human error, a simple typo was made and “123 Main St” was mistakenly entered as “123 Mian St”, it will likely be corrected during the address standardization process.
300
True or False: Hazard resolution is dependent upon the geocoding match level.
What is True. The hazard data resolution can never be higher than the geocoding resolution; so whatever the geocoding resolution is, the hazard data level will be the same or lower.
300
True or False - RMS Medium Term Rates represent RMS' best estimates of hurricane frequencies over the next decade
What is FALSE - RMS Medium Term Rates represent RMS' best estimates of hurricane frequencies over the next five years
400
True or False: Portfolios that are analyzed using aggregate data will always have significantly different results than portfolios that are analyzed using detailed data. A) True – modeling aggregate exposures results in understated loss estimates because the aggregated deductibles are typically very large. B) True – modeling aggregate exposures results in overstated loss estimates because the model uses a conservative assumption when policy-specific data is missing. C) False – the amount of variation in loss results will depend primarily on the resolution of the peril that is being modeled and the spread of the business within a geographic region. D) False – the amount of variation will depend primarily on the number of locations that are in the portfolio (i.e. the more locations, the closer the loss results).
What is What is C) False – the amount of variation in loss results will depend primarily on the resolution of the peril that is being modeled and the spread of the business within a geographic region.
400
A Florida homeowner account submission contains a location that is coded as having wood frame construction with an engineered foundation. The potential data quality issues flagged by this are: A) Data accuracy B) Data validity C) Data bias D) Source data checks E) All of the above
What is What is E) All of the above. Since wood construction single-family homes do not generally have engineered foundations, this would indicate inaccurate data. In addition, this is not an appropriate secondary modifier classification for windstorm exposed risks (but is for commercial earthquake risks). It also brings into question whether other acquired data is relevant for the intended modeling purpose, whether it is biased, and if independent checks of the data are performed.
400
The geocoder’s process of estimating an address’ location on a street segment is called: A) Microzonation B) Extrapolation C) Interpolation D) Area weighting
What is What is C) Interpolation. The geocoder interpolates or inserts the location between the endpoints of a street segment. The result depends upon the address range associated with that segment and the entered address.
400
A building is located at the same latitude and longitude as the centroid of its U.S. ZIP Code. True or False: This building will always generate the same average annual loss estimate for hurricane risk, whether it is geocoded at street address resolution or ZIP Code resolution.
What is False. If the location geocodes at the ZIP Code level, the model will also perform hazard retrieval at that level. If the location geocodes at the street level, the hazard lookup can be much different than at the ZIP Code level, which could impact the AAL.
400
Based on the methodology approved by the Cat Committee, Inland Marine equipment dealers should be modelled with what PCNTCOMPLT?
What is 75%
500
These are the two rules of exposure data analysis.
What are: 1) Understand your end goal 2) Analyze data within a "fitness for purpose" framework
500
Three factors that can cause modelled data to be different from actual exposures.
What are: A) Incorrect data B) Missing data C) Misinterpretation D) Analysis error
500
Why would the geocoder change the ZIP Code of a U.S. address? A) The street address entered could not be found in the ZIP Code entered, but was found in another nearby ZIP Code. B) The software has identified a higher-resolution ZIP Code in the same sorting area. C) The ZIP Code entered is too close to the coast. D) The geocoder should never change the ZIP Code.
What is A) The street address entered could not be found in the ZIP Code entered, but was found in another nearby ZIP Code. There are algorithms built into the software that allow the system to search for likely correct answers. It will then assign each possible answer with a score, ranging from 1 – 10 (1 being the most likely). For example, if there was no “123 Main St” in ZIP Code 33210, but found it in an adjacent ZIP Code, it the software will alter the address entered to include the appropriate ZIP Code.
500
An underwriter in your organization is evaluating a high-value commercial structure for flood coverage. As part of her evaluation, she has entered the location address into a commercially available catastrophe model, and retrieved the site elevation with a street address geocoding match. However, the elevation provided by the model differs by nearly 10 feet (3 meters) from the value obtained by loss control engineers through a site inspection. List at least three potential reasons for this discrepancy.
What are:  Uncertainty in the geocoded location  Uncertainty in the underlying elevation data  Low resolution in the underlying elevation data  Use of different coordinate systems (projection and datum) for measuring location as opposed to elevation  Error in data entry of the address information  Error in calculations for site inspection report  Software or data error in the catastrophe model
500
Describe three of the six adjustments made in the Brit View of Risk
What is a) US WS frequency is assumed to be too high in RMS; blended between RMS and AIR b) Loss Adjustment Expenses - 5% load added for most classes c) Ungeocoded Exposure - loaded to account for these locations d) Demand Surge - loaded in regions not included in Risk Link e) Coastal flooding - explicitly modelled for the Binders book f) Expert Judgement - class-specific adjustments based on actuarial analyses
M
e
n
u