378.CAN SAS REALLY DETECT INSURANCE FRAUD AND CONTROL CLAIM LOSSES USING REAL DATA ANALYTICS?

CAN SAS REALLY DETECT INSURANCE FRAUD AND CONTROL CLAIM LOSSES USING REAL DATA ANALYTICS?

HERE IN THIS PROJECT WE USED THESE SAS STATEMENTS ---DATA STEP | PROC SQL | PROC FREQ | PROC MEANS | PROC UNIVARIATE | MACROS | MDY | INTCK | INTNX | SET | MERGE | APPEND | TRANSPOSE | SAS FUNCTIONS

1.  BUSINESS CONTEXT – Why Insurance Claim Analytics Matters

Insurance companies process millions of claims every year.
Each claim is a potential financial risk.

Insurance analytics teams focus on:

• Detecting fraud
• Reducing claim leakage
• Improving settlement turnaround
• Identifying risky policies
• Ensuring regulatory compliance

Fraud can occur in:

• Health insurance
• Motor insurance
• Property insurance
• Life insurance
• Travel insurance

Even 1–2% fraud can cause crores of loss annually.

This SAS project simulates a real-world claim processing system where:

• Claims arrive
• Risk is calculated
• Fraud is flagged
• Settlement is decided

2. Project Learning Objectives

By completing this project, you will master:

• DATA step programming
• PROC SQL joins & calculations
• Risk scoring logic
• Fraud flagging with macros
• Date difference calculations
• Claim settlement analysis
• Production-style SAS coding

3. Creating Raw Insurance Claims Dataset

data claims_raw;

    format Claim_Date Submission_Date date9.;

    input Claim_ID $ Policy_Type $ Claim_Amount Approval_Time Fraud_Flag$ Settlement_Status $ 

          Claim_Date:date9. Submission_Date:date9.;

    datalines;

C001 Health 120000 5 N Approved 01JAN2024 28DEC2023

C002 Motor 450000 12 Y Pending 05JAN2024 20DEC2023

C003 Travel 60000 3 N Approved 10JAN2024 08JAN2024

C004 Property 950000 25 Y Rejected 12JAN2024 01JAN2024

C005 Health 180000 7 N Approved 15JAN2024 10JAN2024

C006 Motor 320000 15 Y Pending 18JAN2024 02JAN2024

C007 Health 75000 4 N Approved 20JAN2024 17JAN2024

C008 Property 1500000 30 Y Rejected 25JAN2024 05JAN2024

C009 Travel 45000 2 N Approved 28JAN2024 27JAN2024

C010 Motor 280000 10 N Approved 30JAN2024 20JAN2024

C011 Health 220000 9 N Approved 02FEB2024 25JAN2024

C012 Property 680000 18 Y Pending 05FEB2024 15JAN2024

C013 Travel 55000 3 N Approved 07FEB2024 06FEB2024

C014 Motor 410000 14 Y Pending 10FEB2024 28JAN2024

C015 Health 195000 8 N Approved 12FEB2024 05FEB2024

;

run;

proc print data=claims_raw;

run;

OUTPUT:

Obs	Claim_Date	Submission_Date	Claim_ID	Policy_Type	Claim_Amount	Approval_Time	Fraud_Flag	Settlement_Status
1	01JAN2024	28DEC2023	C001	Health	120000	5	N	Approved
2	05JAN2024	20DEC2023	C002	Motor	450000	12	Y	Pending
3	10JAN2024	08JAN2024	C003	Travel	60000	3	N	Approved
4	12JAN2024	01JAN2024	C004	Property	950000	25	Y	Rejected
5	15JAN2024	10JAN2024	C005	Health	180000	7	N	Approved
6	18JAN2024	02JAN2024	C006	Motor	320000	15	Y	Pending
7	20JAN2024	17JAN2024	C007	Health	75000	4	N	Approved
8	25JAN2024	05JAN2024	C008	Property	1500000	30	Y	Rejected
9	28JAN2024	27JAN2024	C009	Travel	45000	2	N	Approved
10	30JAN2024	20JAN2024	C010	Motor	280000	10	N	Approved
11	02FEB2024	25JAN2024	C011	Health	220000	9	N	Approved
12	05FEB2024	15JAN2024	C012	Property	680000	18	Y	Pending
13	07FEB2024	06FEB2024	C013	Travel	55000	3	N	Approved
14	10FEB2024	28JAN2024	C014	Motor	410000	14	Y	Pending
15	12FEB2024	05FEB2024	C015	Health	195000	8	N	Approved

4. Date Calculations using INTCK

data claims_dates;

    set claims_raw;

    Days_To_Submit = intck('day', Submission_Date, Claim_Date);

run;

proc print data=claims_dates;

run;

OUTPUT:

Obs	Claim_Date	Submission_Date	Claim_ID	Policy_Type	Claim_Amount	Approval_Time	Fraud_Flag	Settlement_Status	Days_To_Submit
1	01JAN2024	28DEC2023	C001	Health	120000	5	N	Approved	4
2	05JAN2024	20DEC2023	C002	Motor	450000	12	Y	Pending	16
3	10JAN2024	08JAN2024	C003	Travel	60000	3	N	Approved	2
4	12JAN2024	01JAN2024	C004	Property	950000	25	Y	Rejected	11
5	15JAN2024	10JAN2024	C005	Health	180000	7	N	Approved	5
6	18JAN2024	02JAN2024	C006	Motor	320000	15	Y	Pending	16
7	20JAN2024	17JAN2024	C007	Health	75000	4	N	Approved	3
8	25JAN2024	05JAN2024	C008	Property	1500000	30	Y	Rejected	20
9	28JAN2024	27JAN2024	C009	Travel	45000	2	N	Approved	1
10	30JAN2024	20JAN2024	C010	Motor	280000	10	N	Approved	10
11	02FEB2024	25JAN2024	C011	Health	220000	9	N	Approved	8
12	05FEB2024	15JAN2024	C012	Property	680000	18	Y	Pending	21
13	07FEB2024	06FEB2024	C013	Travel	55000	3	N	Approved	1
14	10FEB2024	28JAN2024	C014	Motor	410000	14	Y	Pending	13
15	12FEB2024	05FEB2024	C015	Health	195000	8	N	Approved	7

claims_dates (INTCK)

• Calculates submission delay, a key fraud indicator.
• Helps identify rushed or delayed claims.
• Used in regulatory turnaround tracking.

5. Risk Score Calculation

data claims_risk;

    set claims_dates;

    Risk_Score = (Claim_Amount/10000) + Approval_Time + (Days_To_Submit*2);

run;

proc print data=claims_risk;

run;

OUTPUT:

Obs	Claim_Date	Submission_Date	Claim_ID	Policy_Type	Claim_Amount	Approval_Time	Fraud_Flag	Settlement_Status	Days_To_Submit	Risk_Score
1	01JAN2024	28DEC2023	C001	Health	120000	5	N	Approved	4	25.0
2	05JAN2024	20DEC2023	C002	Motor	450000	12	Y	Pending	16	89.0
3	10JAN2024	08JAN2024	C003	Travel	60000	3	N	Approved	2	13.0
4	12JAN2024	01JAN2024	C004	Property	950000	25	Y	Rejected	11	142.0
5	15JAN2024	10JAN2024	C005	Health	180000	7	N	Approved	5	35.0
6	18JAN2024	02JAN2024	C006	Motor	320000	15	Y	Pending	16	79.0
7	20JAN2024	17JAN2024	C007	Health	75000	4	N	Approved	3	17.5
8	25JAN2024	05JAN2024	C008	Property	1500000	30	Y	Rejected	20	220.0
9	28JAN2024	27JAN2024	C009	Travel	45000	2	N	Approved	1	8.5
10	30JAN2024	20JAN2024	C010	Motor	280000	10	N	Approved	10	58.0
11	02FEB2024	25JAN2024	C011	Health	220000	9	N	Approved	8	47.0
12	05FEB2024	15JAN2024	C012	Property	680000	18	Y	Pending	21	128.0
13	07FEB2024	06FEB2024	C013	Travel	55000	3	N	Approved	1	10.5
14	10FEB2024	28JAN2024	C014	Motor	410000	14	Y	Pending	13	81.0
15	12FEB2024	05FEB2024	C015	Health	195000	8	N	Approved	7	41.5

claims_risk

• Converts multiple claim factors into one numeric score.
• Enables ranking of claims by financial risk.
• Supports automated fraud logic.

6. Character Cleaning

data claims_clean;

    set claims_risk;

    Policy_Type = propcase(strip(Policy_Type));

    Settlement_Status = upcase(trim(Settlement_Status));

run;

proc print data=claims_clean;

run;

OUTPUT:

Obs	Claim_Date	Submission_Date	Claim_ID	Policy_Type	Claim_Amount	Approval_Time	Fraud_Flag	Settlement_Status	Days_To_Submit	Risk_Score
1	01JAN2024	28DEC2023	C001	Health	120000	5	N	APPROVED	4	25.0
2	05JAN2024	20DEC2023	C002	Motor	450000	12	Y	PENDING	16	89.0
3	10JAN2024	08JAN2024	C003	Travel	60000	3	N	APPROVED	2	13.0
4	12JAN2024	01JAN2024	C004	Property	950000	25	Y	REJECTED	11	142.0
5	15JAN2024	10JAN2024	C005	Health	180000	7	N	APPROVED	5	35.0
6	18JAN2024	02JAN2024	C006	Motor	320000	15	Y	PENDING	16	79.0
7	20JAN2024	17JAN2024	C007	Health	75000	4	N	APPROVED	3	17.5
8	25JAN2024	05JAN2024	C008	Property	1500000	30	Y	REJECTED	20	220.0
9	28JAN2024	27JAN2024	C009	Travel	45000	2	N	APPROVED	1	8.5
10	30JAN2024	20JAN2024	C010	Motor	280000	10	N	APPROVED	10	58.0
11	02FEB2024	25JAN2024	C011	Health	220000	9	N	APPROVED	8	47.0
12	05FEB2024	15JAN2024	C012	Property	680000	18	Y	PENDING	21	128.0
13	07FEB2024	06FEB2024	C013	Travel	55000	3	N	APPROVED	1	10.5
14	10FEB2024	28JAN2024	C014	Motor	410000	14	Y	PENDING	13	81.0
15	12FEB2024	05FEB2024	C015	Health	195000	8	N	APPROVED	7	41.5

claims_clean (Character Functions)

• Standardizes text values for accurate grouping.
• Prevents duplicate categories caused by case or spaces.
• Improves PROC SQL and PROC FREQ reliability.

7. Fraud Detection Macro

%macro fraud_flag(input=, output=);

data &output;

    set &input;

    length Fraud_Category $12;

    if Risk_Score > 80 then Fraud_Category = "HIGH";

    else if Risk_Score > 50 then Fraud_Category = "MEDIUM";

    else Fraud_Category = "LOW";

run;

proc print data=&output;

run;

%mend;

%fraud_flag(input=claims_clean, output=claims_flagged);

OUTPUT:

Obs	Claim_Date	Submission_Date	Claim_ID	Policy_Type	Claim_Amount	Approval_Time	Fraud_Flag	Settlement_Status	Days_To_Submit	Risk_Score	Fraud_Category
1	01JAN2024	28DEC2023	C001	Health	120000	5	N	APPROVED	4	25.0	LOW
2	05JAN2024	20DEC2023	C002	Motor	450000	12	Y	PENDING	16	89.0	HIGH
3	10JAN2024	08JAN2024	C003	Travel	60000	3	N	APPROVED	2	13.0	LOW
4	12JAN2024	01JAN2024	C004	Property	950000	25	Y	REJECTED	11	142.0	HIGH
5	15JAN2024	10JAN2024	C005	Health	180000	7	N	APPROVED	5	35.0	LOW
6	18JAN2024	02JAN2024	C006	Motor	320000	15	Y	PENDING	16	79.0	MEDIUM
7	20JAN2024	17JAN2024	C007	Health	75000	4	N	APPROVED	3	17.5	LOW
8	25JAN2024	05JAN2024	C008	Property	1500000	30	Y	REJECTED	20	220.0	HIGH
9	28JAN2024	27JAN2024	C009	Travel	45000	2	N	APPROVED	1	8.5	LOW
10	30JAN2024	20JAN2024	C010	Motor	280000	10	N	APPROVED	10	58.0	MEDIUM
11	02FEB2024	25JAN2024	C011	Health	220000	9	N	APPROVED	8	47.0	LOW
12	05FEB2024	15JAN2024	C012	Property	680000	18	Y	PENDING	21	128.0	HIGH
13	07FEB2024	06FEB2024	C013	Travel	55000	3	N	APPROVED	1	10.5	LOW
14	10FEB2024	28JAN2024	C014	Motor	410000	14	Y	PENDING	13	81.0	HIGH
15	12FEB2024	05FEB2024	C015	Health	195000	8	N	APPROVED	7	41.5	LOW

fraud_flag Macro

• Automates fraud classification logic.
• Allows consistent risk grading across datasets.
• Supports production-level reuse.

8. PROC FREQ – Fraud Patterns

proc freq data=claims_flagged;

    tables Fraud_Category*Settlement_Status;

run;

OUTPUT:

The FREQ Procedure

Frequency Percent Row Pct Col Pct

Table of Fraud_Category by Settlement_Status Fraud_Category Settlement_Status APPROVED PENDING REJECTED Total HIGH 0 0.00 0.00 0.00 3 20.00 60.00 75.00 2 13.33 40.00 100.00 5 33.33     LOW 8 53.33 100.00 88.89 0 0.00 0.00 0.00 0 0.00 0.00 0.00 8 53.33     MEDIUM 1 6.67 50.00 11.11 1 6.67 50.00 25.00 0 0.00 0.00 0.00 2 13.33     Total 9 60.00 4 26.67 2 13.33 15 100.00

PROC FREQ

• Detects fraud patterns by settlement status.
• Identifies abnormal approval behavior.
• Used in compliance audits.

9. PROC MEANS – Financial Risk

proc means data=claims_flagged mean sum max;

    class Fraud_Category;

    var Claim_Amount Risk_Score;

run;

OUTPUT:

The MEANS Procedure

Fraud_Category	N Obs	Variable	Mean	Sum	Maximum
HIGH	5	Claim_Amount Risk_Score	798000.00 132.0000000	3990000.00 660.0000000	1500000.00 220.0000000
LOW	8	Claim_Amount Risk_Score	118750.00 24.7500000	950000.00 198.0000000	220000.00 47.0000000
MEDIUM	2	Claim_Amount Risk_Score	300000.00 68.5000000	600000.00 137.0000000	320000.00 79.0000000

PROC MEANS

• Summarizes claim exposure by risk level.
• Measures financial impact of fraud.
• Supports management reporting.

10. PROC UNIVARIATE – Risk Distribution

proc univariate data=claims_flagged;

    var Risk_Score;

run;

OUTPUT:

The UNIVARIATE Procedure

Variable: Risk_Score

Moments
N	15	Sum Weights	15
Mean	66.3333333	Sum Observations	995
Std Deviation	59.4452529	Variance	3533.7381
Skewness	1.39401055	Kurtosis	1.90037472
Uncorrected SS	115474	Corrected SS	49472.3333
Coeff Variation	89.6159592	Std Error Mean	15.3486983

Basic Statistical Measures
Location		Variability
Mean	66.33333	Std Deviation	59.44525
Median	47.00000	Variance	3534
Mode	.	Range	211.50000
		Interquartile Range	71.50000

Tests for Location: Mu0=0
Test	Statistic		p Value
Student's t	t	4.321756	Pr > |t|	0.0007
Sign	M	7.5	Pr >= |M|	<.0001
Signed Rank	S	60	Pr >= |S|	<.0001

Quantiles (Definition 5)
Level	Quantile
100% Max	220.0
99%	220.0
95%	220.0
90%	142.0
75% Q3	89.0
50% Median	47.0
25% Q1	17.5
10%	10.5
5%	8.5
1%	8.5
0% Min	8.5

Extreme Observations
Lowest		Highest
Value	Obs	Value	Obs
8.5	9	81	14
10.5	13	89	2
13.0	3	128	12
17.5	7	142	4
25.0	1	220	8

PROC UNIVARIATE

• Analyzes risk score distribution.
• Detects extreme or suspicious values.
• Supports fraud threshold tuning.

11. PROC SQL – High Risk Claims

proc sql;

    create table high_risk as

    select Claim_ID, Policy_Type, Claim_Amount, Risk_Score

    from claims_flagged

    where Risk_Score > 70;

quit;

proc print data=high_risk;

run;

OUTPUT:

Obs	Claim_ID	Policy_Type	Claim_Amount	Risk_Score
1	C002	Motor	450000	89
2	C004	Property	950000	142
3	C006	Motor	320000	79
4	C008	Property	1500000	220
5	C012	Property	680000	128
6	C014	Motor	410000	81

PROC SQL

• Filters and extracts high-risk claims.
• Enables audit-ready fraud listings.
• Supports investigator review.

12. PROC TRANSPOSE – Reporting

proc transpose data=claims_flagged out=claims_transposed;

    by Policy_Type NotSorted;

    id Fraud_Category;

    var Claim_Amount;

run;

proc print data=claims_transposed;

run;

OUTPUT:

Obs	Policy_Type	_NAME_	LOW	HIGH	MEDIUM
1	Health	Claim_Amount	120000	.	.
2	Motor	Claim_Amount	.	450000	.
3	Travel	Claim_Amount	60000	.	.
4	Property	Claim_Amount	.	950000	.
5	Health	Claim_Amount	180000	.	.
6	Motor	Claim_Amount	.	.	320000
7	Health	Claim_Amount	75000	.	.
8	Property	Claim_Amount	.	1500000	.
9	Travel	Claim_Amount	45000	.	.
10	Motor	Claim_Amount	.	.	280000
11	Health	Claim_Amount	220000	.	.
12	Property	Claim_Amount	.	680000	.
13	Travel	Claim_Amount	55000	.	.
14	Motor	Claim_Amount	.	410000	.
15	Health	Claim_Amount	195000	.	.

PROC TRANSPOSE

• Converts risk categories into report-ready columns.
• Used for dashboard and executive summaries.
• Improves visualization structure.

13. PROC APPEND

proc append base=claims_flagged 

            data=high_risk;

run;

proc print data=claims_flagged;

run;

OUTPUT:

Obs	Claim_Date	Submission_Date	Claim_ID	Policy_Type	Claim_Amount	Approval_Time	Fraud_Flag	Settlement_Status	Days_To_Submit	Risk_Score	Fraud_Category
1	01JAN2024	28DEC2023	C001	Health	120000	5	N	APPROVED	4	25.0	LOW
2	05JAN2024	20DEC2023	C002	Motor	450000	12	Y	PENDING	16	89.0	HIGH
3	10JAN2024	08JAN2024	C003	Travel	60000	3	N	APPROVED	2	13.0	LOW
4	12JAN2024	01JAN2024	C004	Property	950000	25	Y	REJECTED	11	142.0	HIGH
5	15JAN2024	10JAN2024	C005	Health	180000	7	N	APPROVED	5	35.0	LOW
6	18JAN2024	02JAN2024	C006	Motor	320000	15	Y	PENDING	16	79.0	MEDIUM
7	20JAN2024	17JAN2024	C007	Health	75000	4	N	APPROVED	3	17.5	LOW
8	25JAN2024	05JAN2024	C008	Property	1500000	30	Y	REJECTED	20	220.0	HIGH
9	28JAN2024	27JAN2024	C009	Travel	45000	2	N	APPROVED	1	8.5	LOW
10	30JAN2024	20JAN2024	C010	Motor	280000	10	N	APPROVED	10	58.0	MEDIUM
11	02FEB2024	25JAN2024	C011	Health	220000	9	N	APPROVED	8	47.0	LOW
12	05FEB2024	15JAN2024	C012	Property	680000	18	Y	PENDING	21	128.0	HIGH
13	07FEB2024	06FEB2024	C013	Travel	55000	3	N	APPROVED	1	10.5	LOW
14	10FEB2024	28JAN2024	C014	Motor	410000	14	Y	PENDING	13	81.0	HIGH
15	12FEB2024	05FEB2024	C015	Health	195000	8	N	APPROVED	7	41.5	LOW
16	.	.	C002	Motor	450000	.			.	89.0
17	.	.	C004	Property	950000	.			.	142.0
18	.	.	C006	Motor	320000	.			.	79.0
19	.	.	C008	Property	1500000	.			.	220.0
20	.	.	C012	Property	680000	.			.	128.0
21	.	.	C014	Motor	410000	.			.	81.0

PROC APPEND

• Merges high-risk cases back into master data.
• Preserves fraud tracking history.
• Used in risk escalation pipelines.

14. PROC DATASETS – Cleanup

proc datasets library=work;

    delete claims_transposed;

quit;

LOG: 

NOTE: Deleting WORK.CLAIMS_TRANSPOSED (memtype=DATA).

PROC DATASETS DELETE

• Removes temporary working tables.
• Reduces memory and workspace clutter.
• Maintains production data hygiene.

15. WHAT THIS PROJECT TEACH US

This is exactly how insurance companies build:

• Fraud detection engines
• Claim risk scoring models
• Settlement decision systems
• Regulatory reports
• Financial risk dashboards

We practiced:

• Real SAS data engineering
• Real business rules
• Real fraud logic
• Real insurance workflows

This project alone is strong enough for:

• Insurance domain interviews
• Banking analytics roles
• SAS Programmer positions
• Risk & compliance jobs

16. CONCLUSION

This project demonstrated how SAS can be used to build a complete insurance claims fraud detection and risk scoring system. By combining DATA steps, PROC SQL, statistical procedures, and macros, we transformed raw claim data into actionable risk intelligence. The model identified high-risk claims, optimized settlement decisions, and improved fraud visibility. This approach mirrors how real insurance companies protect revenue and ensure regulatory compliance.

INTERVIEW QUESTIONS FOR YOU

1.What is the difference between INTCK and INTNX in SAS date handling?

2.How does PROC TRANSPOSE work and when is it used in reporting?

3.What is the role of PROC DATASETS compared to deleting datasets using a DATA step?

About the Author:

SAS Learning Hub is a data analytics and SAS programming platform focused on clinical, financial, and real-world data analysis. The content is created by professionals with academic training in Pharmaceutics and hands-on experience in Base SAS, PROC SQL, Macros, SDTM, and ADaM, providing practical and industry-relevant SAS learning resources.

Disclaimer:

The datasets and analysis in this article are created for educational and demonstration purposes only. They do not represent Insurance Fraud data.

Our Mission:

This blog provides industry-focused SAS programming tutorials and analytics projects covering finance, healthcare, and technology.

This project is suitable for:

SAS Programmer Interviews

SAS Programmer Job Seekers

SAS Analysts

Follow Us On : 

 

--->Follow our blog for more SAS-based analytics projects and industry data models.

To deepen your understanding of SAS analytics, please refer to our other data science and industry-focused projects listed below:

1.CAN SAS ANALYZE A FISH DATASET USING BASIC PROCS AND MACROS?

2.How to learn retain and sum statements in sas?

3.CAN SAS SIMPLY MANAGE AND ANALYZE A GLOBAL AUDI AQUARIUM FISH INVENTORY?

About Us | Contact | Privacy Policy