Managing Data for Analytics. April 14, 2015

Transcription

1 Managing Data for Analytics April 14,

2 Importance of Predictive Analytics Predictive Analytics can help insurers be more effective in all segments of the value chain Marketing Target and acquire the right customers Actuarial Prices that accurately reflect risk Underwriting Select the proper risks and proper products Claims Identify suspicious claims The industry is getting more competitive Top 10 personal auto insurers had 1/2 the market share in 1980; now they have 2/3 of the market share Only the fittest will survive; analytics can provide the needed competitive advantage The industry has recognized the value of analytics 2

3 Who Uses Predictive Modeling? Predictive analytics is used most often in personal lines. Use of predictive analytics by size of personal lines book 100% of the larger personal lines insurers we surveyed use predictive analytics! Of course, personal lines (and PL auto, in particular) is the largest and one of the most competitive segments of the P&C market. Insurers are looking for any competitive edge they can find. Source: ISO and Earnix 2013 Insurance Predictive Modeling Survey

4 How Insurers Use Predictive Modeling Pricing is the most common use of predictive modeling. Predictive modeling use by function A majority of insurers also use predictive modeling for underwriting at least frequently. 100% 90% 80% 70% 60% 4% 13% 39% 9% 21% But there is still significant usage in marketing, claims, and reserving. 50% 40% 30% 20% 10% 0% 42% 32% 20% 12% 14% 14% 17% 18% 11% 18% 14% 9% 9% 7% 9% Rarely Occasionally Frequently Always Source: ISO and Earnix 2013 Insurance Predictive Modeling Survey

5 Predictive Modeling Challenges Lack of sufficient data is the biggest challenge both quantity and scope. Predictive modeling challenges Lack of skilled modelers is a close second most challenging factor for those building an internal predictive modeling capability. Lack of sufficient data Not enough skilled modelers Need additional computing resources Need better modeling tools 23% 31% 47% 53% We have no challenges 6% Other 9% Source: ISO and Earnix 2013 Insurance Predictive Modeling Survey

6 Data Challenges Data is a challenge for everybody, but large and small insurers have different challenges. Larger insurers are most concerned with data quality. Smaller insurers don t have enough observations. 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Data challenges by company size 3% 30% 7% 18% 42% <$1B 13% 57% 6% 6% 17% >$1B Other Data is not clean, tough to use Data is not current Not enough variability in the data Not enough observations Numbers may not add up to 100% due to rounding Source: ISO and Earnix 2013 Insurance Predictive Modeling Survey

7 Third-Party Data More than 90% of insurers supplement their internal data with one or more types of third-party data. The most common data types are credit-related data and geo-demographic data. Types of third-party data used Insurance score or raw credit attributes Geo-demographical data Competitive pricing data 53% 67% 80% Catastrophe models data 46% Weather data 42% 3rd party telematics data 9% Other 5% Source: ISO and Earnix 2013 Insurance Predictive Modeling Survey

8 Data Preparation Preparing data for analysis is a major bottleneck and drain on resources for most insurers. Data extraction and preparation time 37% 54% of insurers typically spend more than 3 months to prepare their data for a project. 18% 28% 17% Less than 1 month 1 2 months 3 6 months More than 6 months Source: ISO and Earnix 2013 Insurance Predictive Modeling Survey

9 Good Data: Good Analytics Good quality data can often compensate for mediocre analysis but, the reverse is never true. No matter how skilled the analyst, bad data will always lead to bad results! 9

10 Data Use for Analytics is Different Some Characteristics of Analytics Use of Data Sophisticated Users Ad Hoc & Iterative Repurposed Data Granular and Denormalized Data Data Quality and Metadata are different 10

11 Sophisticated Users Most modelers will have advanced degrees in Statistics, Economics, Applied Mathematics, etc. Users will be looking at the data from new perspectives and using the data in new ways Can lead to new insights into the data for data owners can also cause friction. Data managers need to understand the predictive analytics process 11

12 Predictive Analytics Process Overview Business Understanding Data Understanding Data Preparation Deployment Data Modeling Evaluation CRISP DM: Cross-Industry Standard Process for Data Mining 12

13 Ad Hoc Nature of Data Access Analysts will design their queries as needed The iterative nature of the analytic process means the analyst will be back again and again for more and different data There is no standard analytics data report that can be specified when designing the data resource. Structure needs to efficiently and flexibly support ad hoc quiries. 13

14 Repurposed Data Rarely are the operational data stores collected into a single Enterprise Data Warehouse You will need to create a useful analytic data store Even more rare, is data that has been collected specifically for analytics usually, analytics is an opportunistic user of data that has been collected for other purposes Data will need to be cleaned, transformed, conformed, and documented before it is certified as fit for use for analytics and included in the analytic data store 14

15 Insurance Company Data Sources Insurance companies collect vast quantities of data in the course of business Typical Insurance Analytics Data Sources Customer Relationship Management Quoting/Underwriting Policy Management Billing Claims Audit Actuarial Research Financial Reporting Publically Available Data Third-party data vendors 15

16 Granular and Denormalized Data End goal Always remember the goal Two-dimensional flat file for input into modeling software Each record contains an identifier, candidate predictor variables for testing, and one or more target variables Analysis requires historical data and the vintage of the predictor variables must be matched to the target variables Must support the granularity required for level analysis 16

17 Granularity What does each record represent? Common record types for insurance analytics Customer-related First named Insured Household Quote Policy Coverage Claim Geography Census tract County State Underwriting Territory Zip Code 17

18 Implications for Analytic Database Design Star Schema is often adopted to support analytics Data will often be denormalized and aggregated from source systems Analytic databases will often grow to contain more history than the source systems. Plan for growth. Every variable needs a vintage Indexing needs will be imperfectly defined. Count on supporting multiple table joins from any direction = many indexes. Granularity pick the lowest level as your base This means more data, but it is the most flexible design. Data can usually be aggregated to a higher level of granularity but you can never go below your base level. 18

19 Data Quality and Metadata Important Data Quality measures for Analytics: Accuracy Reliability Timeliness Completeness Availability Permissibility Analysts usually can t control the quality of the data when acquired. So, they must at least know the quality of the data in order to determine the usefulness of the data. Metadata documentation of this information 19

20 Data Quality: Accuracy Accuracy - How well does the data element describe the object or fact in the real world? Too much detail just adds unwanted noise Example: a person s height expressed in millimeters Too little detail is not useful Example: a person s height expressed in meters Sometimes the measure is just not meaningful to your application Example: a person s height expressed as a ratio of his income 20

21 Data Quality: Reliability Reliability - Is the measurement repeatable and consistent for the object or fact in the real world? Example: How was the time required to complete a customer call measured? Automatically and electronically by computer By a third party using a stop watch Self reported by operator using a wrist watch Recalled by customer in a survey 21

22 Data Quality: Timeliness Timeliness - Is the data measured at the time that a prediction would have been made? Example: A results of a loss control visit can be very predictive of the severity of a commercial property loss. However, that data won t be useful for a quoting model if the loss control visits only occur after a policy has been has been written. 22

23 Data Quality: Completeness Completeness - Is the data available for the vast majority of the cases in the database? Whether a policyholder has a fire extinguisher available in the home may be a great predictor of fire severity, but what if that data is only available for 0.5% of policyholders? 23

24 Data Quality: Availability Availability - Will the data be available in the appropriate form and updated at appropriate intervals to meet business needs? Example: Survey data It s sometimes available to build a model but often not available when it comes time to make a prediction. 24

25 Data Quality: Permissibility Permissibility - Will use of the data be legal and comply with regulations? Example: Gender is allowed to be used for rating personal auto insurance in some states but not in others. 25

26 Metadata Metadata has many definitions we mean information about the data that the analyst needs to know in order to use the data appropriately Analytic Metadata Needs: Owner/Source Restrictions on use Vintage/update frequency/amount of history Summary statistics Data quality metrics 26

27 New Product: ISO RiskElements Provide archived data to P&C insurers to use as a significant source of the raw material for their in-house predictive modeling efforts. ISO will build on and leverage its experience managing insurance data and assembling analytic datasets as well as its unique data assets. Goal Become the first and primary source of data for predictive modeling projects for the P&C insurance industry. 27

28 RiskElements Product Design RiskElements Data Asset Vintage: Delivery Modes Verisk Data Data appended to customersupplied file Public Data 3 rd Party Proprietary Data Analytic file built from scratch 28

29 Conclusion Advanced analytics requires different data management support than most other uses of the data Two broad areas that demonstrate those needs: Database design Data quality and metadata Strive to build an analytic data store that considers the unique needs of analytics. 29

30 Questions? Phil Hatfield, J.D., CPCU Head of Modeling Data Services for ISO No part of this presentation may be copied or redistributed without the prior written consent of ISO. This material was used exclusively as an exhibit to an oral presentation. It may not be, nor should it be relied upon as reflecting, a complete record of the discussion..