Dataset Description: Insurance Claims Prediction

Introduction: In the insurance industry, accurately predicting the likelihood of claims is essential for risk assessment and policy pricing. However, insurance claims datasets frequently suffer from class imbalance, where the number of non-claims instances far exceeds that of actual claims. This class imbalance poses challenges for predictive modeling, often leading to biased models favoring the majority class, resulting in subpar performance for the minority class, which is typically of greater interest.

Dataset Overview: The dataset utilized in this project comprises historical data on insurance claims, encompassing a variety of information about the policyholders, their demographics, past claim history, and other pertinent features. The dataset is structured to facilitate predictive modeling tasks aimed at accurately identifying the likelihood of future insurance claims.

Key Features: 1. Policyholder Information: This includes demographic details such as age, gender, occupation, marital status, and geographical location. 2. Claim History: Information regarding past insurance claims, including claim amounts, types of claims (e.g., medical, automobile), frequency of claims, and claim durations. 3. Policy Details: Details about the insurance policies held by the policyholders, such as coverage type, policy duration, premium amount, and deductibles. 4. Risk Factors: Variables indicating potential risk factors associated with policyholders, such as credit score, driving record (for automobile insurance), health status (for medical insurance), and property characteristics (for home insurance). 5. External Factors: Factors external to the policyholders that may influence claim likelihood, such as economic indicators, weather conditions, and regulatory changes.

Objective: The primary objective of utilizing this dataset is to develop robust predictive models capable of accurately assessing the likelihood of insurance claims. By leveraging advanced machine learning techniques, such as classification algorithms and ensemble methods, the aim is to mitigate the effects of class imbalance and produce models that demonstrate high predictive performance across both majority and minority classes.

Application Areas: 1. Risk Assessment: Assessing the risk associated with insuring a particular policyholder based on their characteristics and historical claim behavior. 2. Policy Pricing: Determining appropriate premium amounts for insurance policies by estimating the expected claim frequency and severity. 3. Fraud Detection: Identifying fraudulent insurance claims by detecting anomalous patterns in claim submissions and policyholder behavior. 4. Customer Segmentation: Segmenting policyholders into distinct groups based on their risk profiles and insurance needs to tailor marketing strategies and policy offerings.

Conclusion: The insurance claims dataset serves as a valuable resource for developing predictive models aimed at enhancing risk management, policy pricing, and overall operational efficiency within the insurance industry. By addressing the challenges posed by class imbalance and leveraging the rich array of features available, organizations can gain valuable insights into insurance claim likelihood and make informed decisions to mitigate risk and optimize business outcomes.

Context

The company has shared its annual car insurance data. Now, you have to find out the real customer behaviors over the data.

Content

The columns are resembling practical world features. The outcome column indicates 1 if a customer has claimed his/her loan else 0. The data has 19 features from there 18 of them are corresponding logs which were taken by the company.

Acknowledgements

Mostly the data is real and some part of it is also generated by me.

Inspiration

The data is so well balanced that it will help kagglers find a better intuition of real customers and find the deepest story lien within it.

The data is formatted as a spreadsheet, encompassing the primary activities over a span of three full years (November 2015 to December 2018) concerning non-life motor insurance portfolio. This dataset comprises 105,555 rows and 30 columns. Each row signifies a policy transaction, while each column represents a distinct variable.

The dataset, named "insurance_claims.csv", is a comprehensive collection of insurance claim records. Each row represents an individual claim, and the columns represent various features associated with that claim.

The dataset is, highlighting features like 'months_as_customer', 'age', policy_number, ...etc. The main focus is the 'fraud_reported' variable, which indicates claim legitimacy.

Claims data were sourced from various insurance providers, encompassing a diverse array of insurance types including vehicular, property, and personal injury. Each claim's record provides an in-depth look into the individual's background, claim specifics, associated documentation, and feedback from insurance professionals.

The dataset further includes specific indicators and parameters that were considered during the claim's assessment, offering a granular look into the complexities of each claim.

For privacy reasons, and in agreement with the participating insurance providers, certain personal details and specific identifiers have been anonymized. Instead of names or direct identifiers, each entry is associated with a unique ID, ensuring data privacy while retaining data integrity.

The insurance claims were subjected to rigorous examination, encompassing both manual assessments and automated checks. The end result of this examination, specifically whether a claim was deemed fraudulent or not, is clearly indicated for each record.

Autobi(Automobile Bodily Injury Claims) -

The data contains information on demographic information about the claimant, attorney involvement and the economic loss (LOSS, in thousands), among other variables.The full data contains over 70,000 closed claims based on data from thirty-two insurers.

A data frame with 1340 observations on the following 8 variables.

CASENUM- Case number to identify the claim, a numeric vector ATTORNEY- Whether the claimant is represented by an attorney (=1 if yes and =2 if no), a numeric vector CLMSEX - Claimant's gender (=1 if male and =2 if female), a numeric vector MARITAL- claimant's marital status (=1 if married, =2 if single, =3 if widowed, and =4 if divorced/separated), a numeric vector CLMINSUR- Whether or not the driver of the claimant's vehicle was uninsured (=1 if yes, =2 if no, and =3 if not applicable), a numeric vector SEATBELT- Whether or not the claimant was wearing a seatbelt/child restraint (=1 if yes, =2 if no, and =3 if not applicable), a numeric vector CLMAGE- Claimant's age, a numeric vector LOSS- The claimant's total economic loss (in thousands), a numeric vector

AutoClaims(Automobile Insurance Claims) -

A data frame with 6773 observations on the following 5 variables.

STATE CLASS - Rating class of operator, based on age, gender, marital status, use of vehicle GENDER AGE - Age of operator PAID - Amount paid to settle and close a claim

AutoCollision(Automobile UK Collision Claims)

8,942 collision losses from private passenger United Kingdom (UK) automobile insurance policies. The average severity is in pounds sterling adjusted for inflation.

A data frame with 32 observations on the following 4 variables.

Age - Age of driver Vehicle_Use - Purpose of the vehicle use Severity - Average amount of claims Claim_Count - Number of claims

Additional information can be found in the document: https://cran.r-project.org/web/packages/insuranceData/index.html

Data provided by insurers, on the premiums written and claims incurred for the 2013 fiscal year. Based on reporting on the consolidated pages of the P&C-1 or Life-1 Annual returns. This data is also reported in the Superintendent of Insurance’s Annual Report.

Dataset

This dataset was created by xiaomengsun

Contents

Louisiana had the most expensive annual car insurance premiums at ***** U.S. dollars for full coverage. Alaska ranked in first place, having the highest annual cost for minimum car insurance coverage at *** U.S. dollars.Why it varies state by state The huge variance in premiums between states is due to the difference in state laws, the percentage of uninsured drivers in the state, the frequency of natural disasters, and claim rates. For instance, Michigan has a no-fault car insurance system, which means that claims are more common. This drives up the cost of insurance for all drivers because insurers need to pay out more money in claims. Male drivers also pay more There is also a difference between premiums among different age groups. In 2025, 25-year-old male drivers paid more per month than 25-year-old female drivers did. This is due to the higher incidence of accidents among young male drivers. This means that young drivers in states that already have higher premiums must pay a lot for car insurance.

The frequency of private passenger comprehensive auto insurance claims for physical damage in the United States rose to **** per 100 car years in 2023, compared to *** in 2020. This was the highest frequency recorded over the past 15 years.

This dataset comprises 9,134 records of auto insurance claims, encompassing a broad range of attributes related to customer profiles and policy details. Key columns include demographic information such as Customer, State, Gender, Income, and Education, along with policy-specific data like Coverage, Policy Type, and Monthly Premium Auto. This dataset also contains indices for various categorical attributes, including Coverage Index, Education Index, and Vehicle Class Index, which facilitate the quantification of qualitative information. Additionally, the dataset tracks metrics related to policy performance and customer interaction, such as the Number of Open Complaints, Months Since Last Claim, and Total Claim Amount.

To provide a comprehensive view of the insurance landscape, the dataset includes detailed attributes about policy effectiveness and customer engagement. Features such as Effective To Date, Renew Offer Type, Sales Channel, and Vehicle Size contribute to understanding how different factors impact insurance claims. This rich dataset offers valuable insights into customer behavior, policy performance, and overall claim dynamics, making it a robust resource for analyzing trends and patterns in auto insurance claims.

This dataset was initially created in 2011 with values in 2011 dollars. To reflect current economic conditions, I updated it to 2024 dollars using a factor provided by ChatGPT. Additionally, I incorporated index columns to facilitate research and analysis.

In 2022, there were more than ************* auto insurance claims submitted in Germany. The largest share was for comprehensive, or Vollkasko, insurance, which accounted for *** million claims, followed by third-party liability with **** million claims.

Here are a few use cases for this project:

1. Insurance Assessment: This model can be used by insurance companies to automate the process of assessing car damage in insurance claims. By simply using photographs of the damaged vehicle, the model can identify the type and extent of damage, making the claim processing faster and more objective.

2. Automotive Repair Estimates: Car repair shops can use this model to get an approximate idea of the damage and therefore provide a more accurate cost estimate for their clients. It can also assist in identifying nonobvious damage.

3. Used Car Market Evaluation: This model can be used in used car platforms to evaluate the current condition of the cars listed for sale. By identifying existing damage, buyers can make more informed decisions and sellers can price their vehicles more accurately.

4. Law Enforcement and Road Safety: Traffic police and accident investigation teams can utilize this model to evaluate the types of damages after a road accident. It will assist in rebuilding the accident scenario, providing insights during investigations.

5. Auto-manufacturing Quality Control: Automobile manufacturers can use this model in their factories to automatically inspect new cars for any damage or misaligned/missing parts before they are dispatched from the factory, ensuring quality control.

State Farm Mutual Automobile Insurance was the leading private passenger car insurer in the United States in 2024, with premiums written amounting to approximately 68 billion U.S. dollars. Progressive Corporation, and Berkshire Hathaway Inc. were the next largest insurers in this sector. State Farm: a background State Farm Mutual Automobile Insurance was founded in 1922 and is headquartered in Bloomington, Illinois. In 2024, the insurer was the largest writer of property and casualty insurance in the United States. They provide vehicle, homeowners, renters, life and annuities, health, disability and flood insurance among several other insurance products. Net promoter score and ad spend of State Farm Despite their market leader status, State Farm's net promoter score puts them in the middle of the pack, with only 42 percent of their customers saying they would recommend the insurer. However, their nearest competitors did not score any better, with Progressive receiving a NPS of only 38 percent in the same analysis. The three largest car insurers were also the biggest spenders on advertising.

Description: This dataset contains 1,000 rows of synthetic data simulating car insurance premiums, calculated using a linear formula. It incorporates key features such as driver age, driving experience, accident history, annual mileage, and car manufacturing year to predict the insurance premium. The dataset is ideal for exploring linear regression models, feature importance analysis, and predictive modeling in the insurance industry. It was inspired by real-world factors influencing insurance premiums, ensuring realistic patterns and meaningful insights.

Dataset from Singapore Department of Statistics. For more information, visit https://data.gov.sg/datasets/d_abcfd12381e7f8d175280d999cdb2dea/view

Description This dataset and project are part of ClaimWise AI, an intelligent automation service designed to streamline auto insurance claim processing. All data in this release was collected and curated by our team, ensuring originality and alignment with real-world claim processing scenarios.

What’s inside

• claim_metadata.xlsx: Structured metadata of insurance claims including claim IDs, incident details, vehicle attributes, and fraud indicators.
• Auto-claim service pipeline: Demonstrates how to leverage machine learning, embeddings, and multimodal AI to process claim data and simulate real-world automation.

Note on Images The pipeline references car crash and accident images as part of embedding and similarity checks. These images were also collected by our team from publicly available resources and curated for research purposes. They are not redistributed in this dataset but are used internally to illustrate how ClaimWise AI can handle multimodal data.

Key Features

• Automates claim intake and metadata standardization.
• Uses AI embeddings for fraud detection and duplicate claim checks.
• Supports integration with curated external crash images for damage assessment.
• Provides an end-to-end demo of an AI-powered insurance claim automation workflow.

Use Cases

• Fraud detection in auto claims
• Accident severity classification
• Claim similarity and duplicate detection
• AI-enabled insurance process optimization

The DFS ranks automobile insurance companies doing business in New York State based on the number of consumer complaints upheld against them as a percentage of their total business over a two-year period. Complaints typically involve issues like delays in the payment of no-fault claims and nonrenewal of policies. Insurers with the fewest upheld complaints per million dollars of premiums appear at the top of the list. Those with the highest complaint ratios are ranked at the bottom.

This file contains ultimate claims data taken from the private motor National Claims Iinformation Database (NCID). The claims are grouped together by accident year, the year in which the accident occurred. Not all claims are paid in the lifetime of the policy. Some claims, injury claims in particular, can take many years to be settled and be fully paid. Insurers estimate the cost/number of claims expected for a particular accident year, and this known as the ultimate cost/number of claims. The ultimate cost/number of claims is recalculated regularly, based on the most up-to-date information available. The more time that has passed since the accident year, the more certain the ultimate cost of claims becomes. To view the detailed NCID report kindly refer to the centralbank publication link in the Landing Page section under Additional Info.

Switzerland Non Life Insurance: Claims Paid: Liability and Motor data was reported at 4,676.000 CHF mn in 2016. This records a decrease from the previous number of 4,802.000 CHF mn for 2015. Switzerland Non Life Insurance: Claims Paid: Liability and Motor data is updated yearly, averaging 4,628.000 CHF mn from Dec 2000 (Median) to 2016, with 17 observations. The data reached an all-time high of 4,918.000 CHF mn in 2009 and a record low of 3,844.000 CHF mn in 2000. Switzerland Non Life Insurance: Claims Paid: Liability and Motor data remains active status in CEIC and is reported by Swiss Financial Market Supervisory Authority. The data is categorized under Global Database’s Switzerland – Table CH.RG011: Non Life Insurance: Claims Paid.

Here are a few use cases for this project:

1. Insurance Claims Processing: This model can be used by auto insurance companies to automate the process of assessing the damage on vehicles involved in accidents. After a collision, policyholders can take pictures of their damaged vehicles, and the model can identify and classify the damage, speeding up the claims process.

2. Vehicle Repair Estimates: Auto repair shops can utilize this model to quickly generate estimates for vehicle repairs. By using it to catalogue damage, repairs needed for specific parts could be more accurately priced.

3. Online Car Selling Platforms: This model could be used on platforms where used cars are sold. Sellers could upload images of their cars, and the model could assess any visible damage, providing potential buyers with more information about the condition of the vehicle.

4. Traffic Accident Analysis: Law enforcement or accident investigators could use this model to help determine the sequence of events in a car accident. By identifying the damaged parts of the vehicles involved, it could offer clues to how the accident happened.

5. Car Rental Services: Car Rental companies could implement this model to automatically evaluate the condition of cars when they are returned by customers. This could identify any new damages or irregularities as compared to the car's condition at the start of the rental period.