Broken Promises, Corrupted Records & Divorce Data Engineering Excellence

Why Divorce Data Requires Special Cleaning

Divorce registries are among the most complex administrative datasets because they contain:

Former spouse information
Court case details
Legal settlement amounts
Custody information
Regional court jurisdictions
Filing dates
Final judgment dates
Lawyer information

Small errors can produce:

Incorrect legal reports
Wrong custody statistics
Duplicate court cases
Invalid settlement calculations
Misleading demographic analyses

1. Create a Raw Divorce Dataset in SAS

SAS Raw Divorce Dataset

data divorce_raw;

length Case_ID $12 Husband_Name $40 Wife_Name $40 Court_Code $8 Lawyer_Email $60

Settlement_Amount $15 Custody_Status $20 Case_Status $15 Filing_Date_Raw $20

Final_Order_Raw $20 City $25;

input Case_ID $ Husband_Name $ Wife_Name $ Years_Married Children_Count

Court_Code $ Lawyer_Email $ Settlement_Amount $ Custody_Status $ Case_Status $

Filing_Date_Raw $ Final_Order_Raw $ City $;

datalines;

D1001 Ravi Anjali 12 2 AP01 lawyer1@court.in 500000 Joint Closed 12-01-2025 15-04-2025 Hyderabad

D1002 Ravi Anjali 12 2 AP01 lawyer1@court.in 500000 Joint Closed 12-01-2025 15-04-2025 Hyderabad

D1003 Kiran Meena -2 1 TN01 invalidmail 700000 Mother Active 15/02/2025 . Chennai

D1004 Suresh Kavya 45 0 KA01 lawyer4@court.in -800000 Father Closed 18MAR2025 20APR2025 Bangalore

D1005 Mohan Radha 8 3 AP02 lawyer5@court.in abc500 Joint Active . . Vijayawada

D1006 Akash Nisha 110 1 MH01 lawyer6@court.in 900000 Mother Closed 12APR2025 15MAY2025 Mumbai

D1007 Arjun Priya 15 2 XX01 lawyer7@court.in 300000 Joint Closed 10-10-2025 12-12-2025 Delhi

D1008 Tarun Pooja 7 . AP03 lawyer8@court.in 450000 Father Active 20JAN2025 . Hyderabad

D1009 Raj Divya 14 2 KA02 lawyer9@court.in 1000000 Joint Closed 15FEB2025 18MAR2025 Mysore

D1010 Naveen Sneha 5 1 AP04 lawyer10@court.in 550000 Mother Active 17MAR2025 . Vizag

D1011 Rakesh Simran 20 2 MH02 lawyer11@court.in 850000 Joint Closed 18APR2025 22JUN2025 Pune

D1012 Ajay Priti 18 1 TN02 lawyer12@court.in 620000 Father Closed 10MAY2025 20JUL2025 Chennai

D1013 Deepak Rani 13 2 AP05 lawyer13@court.in 580000 Joint Active 14JUN2025 . Guntur

D1014 Vinod Rekha 9 0 KA03 lawyer14@court.in 400000 Mother Closed 15JUL2025 18SEP2025 Hubli

D1015 Rahul Keerthi 16 2 AP06 lawyer15@court.in 760000 Joint Active 21AUG2025 . Tirupati

D1016 Mahesh Swathi 4 1 TN03 lawyer16@court.in 250000 Father Closed 12SEP2025 15NOV2025 Madurai

D1017 Dinesh Latha 11 3 MH03 lawyer17@court.in 690000 Joint Active 17OCT2025 . Nagpur

D1018 Harish Deepa 6 0 AP07 lawyer18@court.in 350000 Mother Closed 20NOV2025 22DEC2025 Nellore

D1019 Vijay Sita 21 2 KA04 lawyer19@court.in 920000 Joint Closed 05DEC2025 10JAN2026 Belgaum

D1020 Anand Jyothi 10 1 AP08 lawyer20@court.in 480000 Father Active 12DEC2025 . Kurnool

;

run;

proc print data = divorce_raw;

run;

OUTPUT:


Obs	Case_ID	Husband_Name	Wife_Name	Court_Code	Lawyer_Email	Settlement_Amount	Custody_Status	Case_Status	Filing_Date_Raw	Final_Order_Raw	City	Years_Married	Children_Count
1	D1001	Ravi	Anjali	AP01	lawyer1@court.in	500000	Joint	Closed	12-01-2025	15-04-2025	Hyderabad	12	2
2	D1002	Ravi	Anjali	AP01	lawyer1@court.in	500000	Joint	Closed	12-01-2025	15-04-2025	Hyderabad	12	2
3	D1003	Kiran	Meena	TN01	invalidmail	700000	Mother	Active	15/02/2025		Chennai	-2	1
4	D1004	Suresh	Kavya	KA01	lawyer4@court.in	-800000	Father	Closed	18MAR2025	20APR2025	Bangalore	45	0
5	D1005	Mohan	Radha	AP02	lawyer5@court.in	abc500	Joint	Active			Vijayawada	8	3
6	D1006	Akash	Nisha	MH01	lawyer6@court.in	900000	Mother	Closed	12APR2025	15MAY2025	Mumbai	110	1
7	D1007	Arjun	Priya	XX01	lawyer7@court.in	300000	Joint	Closed	10-10-2025	12-12-2025	Delhi	15	2
8	D1008	Tarun	Pooja	AP03	lawyer8@court.in	450000	Father	Active	20JAN2025		Hyderabad	7	.
9	D1009	Raj	Divya	KA02	lawyer9@court.in	1000000	Joint	Closed	15FEB2025	18MAR2025	Mysore	14	2
10	D1010	Naveen	Sneha	AP04	lawyer10@court.in	550000	Mother	Active	17MAR2025		Vizag	5	1
11	D1011	Rakesh	Simran	MH02	lawyer11@court.in	850000	Joint	Closed	18APR2025	22JUN2025	Pune	20	2
12	D1012	Ajay	Priti	TN02	lawyer12@court.in	620000	Father	Closed	10MAY2025	20JUL2025	Chennai	18	1
13	D1013	Deepak	Rani	AP05	lawyer13@court.in	580000	Joint	Active	14JUN2025		Guntur	13	2
14	D1014	Vinod	Rekha	KA03	lawyer14@court.in	400000	Mother	Closed	15JUL2025	18SEP2025	Hubli	9	0
15	D1015	Rahul	Keerthi	AP06	lawyer15@court.in	760000	Joint	Active	21AUG2025		Tirupati	16	2
16	D1016	Mahesh	Swathi	TN03	lawyer16@court.in	250000	Father	Closed	12SEP2025	15NOV2025	Madurai	4	1
17	D1017	Dinesh	Latha	MH03	lawyer17@court.in	690000	Joint	Active	17OCT2025		Nagpur	11	3
18	D1018	Harish	Deepa	AP07	lawyer18@court.in	350000	Mother	Closed	20NOV2025	22DEC2025	Nellore	6	0
19	D1019	Vijay	Sita	KA04	lawyer19@court.in	920000	Joint	Closed	05DEC2025	10JAN2026	Belgaum	21	2
20	D1020	Anand	Jyothi	AP08	lawyer20@court.in	480000	Father	Active	12DEC2025		Kurnool	10	1

Data Quality Problems Intentionally Added

Issue	Example
Duplicate Cases	D1001, D1002
Invalid Marriage Years	-2
Unrealistic Marriage Duration	110
Invalid Court Code	XX01
Invalid Email	invalidmail
Character Amount	abc500
Negative Settlement	-800000
Missing Children Count	.
Missing Dates	.

These are realistic issues commonly found in legal and government datasets.

SAS Cleaning Using DIFFERENT Statements

Instead of ARRAYS and COMPRESS, use:

VERIFY()
TRANWRD()
CATX()
FIND()
INPUTN()
COALESCEC()
MISSING()
NOTDIGIT()

Explanation

This SAS program creates the initial raw divorce registration dataset named DIVORCE_RAW. In any real-world data engineering project, the first step is usually to build or receive a source dataset from external systems such as government courts, legal departments, CRM systems, or data warehouses. This dataset intentionally contains quality issues that will later be identified and corrected through data cleaning procedures.

The DATA statement tells SAS to create a new dataset called DIVORCE_RAW. Every observation entered through the DATALINES statement becomes one row in the dataset.

The LENGTH statement is extremely important because it explicitly defines the storage size of character variables. For example, Case_ID $12 reserves twelve characters for the case identifier. If the LENGTH statement is omitted, SAS determines character lengths based on the first encountered value, which may lead to truncation problems later. Production programmers almost always define LENGTH before reading data.

Variables such as Husband_Name, Wife_Name, Court_Code, Lawyer_Email, and City are defined as character variables using the $ symbol. Numeric variables like Years_Married and Children_Count do not require $.

The INPUT statement tells SAS how to read each field. The ampersand (&) modifier allows SAS to read character values that may contain embedded spaces. This is useful when names contain multiple words.

Finally, the DATALINES statement provides the actual divorce records. These records intentionally contain duplicates, invalid emails, unrealistic marriage durations, and inconsistent values to simulate real-world legal data challenges. This raw dataset serves as the foundation for all subsequent cleaning, validation, reporting, and analytical activities.

2.SAS Cleaning Program

data divorce_clean;

set divorce_raw;

Lawyer_Email=lowcase(strip(Lawyer_Email));

Court_Code=upcase(Court_Code);

City=propcase(City);

if Years_Married < 0 then Years_Married=.;

if Years_Married > 60 then Years_Married=.;

Settlement_Num=input(Settlement_Amount,?? best12.);

Settlement_Num=abs(Settlement_Num);

if missing(Settlement_Num) then

Settlement_Flag='INVALID';

else Settlement_Flag='VALID';

if find(Lawyer_Email,'@')=0 then

Email_Flag='INVALID';

else Email_Flag='VALID';

Filing_Date=input(Filing_Date_Raw,anydtdte20.);

Final_Order=input(Final_Order_Raw,anydtdte20.);

format Filing_Date Final_Order date9.;

Case_Label=catx('-',Court_Code,Case_ID,City);

drop Settlement_Amount;

rename Settlement_Num = Settlement_Amount;

run;

proc print data = divorce_clean;

run;

OUTPUT:


Obs	Case_ID	Husband_Name	Wife_Name	Court_Code	Lawyer_Email	Custody_Status	Case_Status	Filing_Date_Raw	Final_Order_Raw	City	Years_Married	Children_Count	Settlement_Amount	Settlement_Flag	Email_Flag	Filing_Date	Final_Order	Case_Label
1	D1001	Ravi	Anjali	AP01	lawyer1@court.in	Joint	Closed	12-01-2025	15-04-2025	Hyderabad	12	2	500000	VALID	VALID	01DEC2025	15APR2025	AP01-D1001-Hyderabad
2	D1002	Ravi	Anjali	AP01	lawyer1@court.in	Joint	Closed	12-01-2025	15-04-2025	Hyderabad	12	2	500000	VALID	VALID	01DEC2025	15APR2025	AP01-D1002-Hyderabad
3	D1003	Kiran	Meena	TN01	invalidmail	Mother	Active	15/02/2025		Chennai	.	1	700000	VALID	INVALID	15FEB2025	.	TN01-D1003-Chennai
4	D1004	Suresh	Kavya	KA01	lawyer4@court.in	Father	Closed	18MAR2025	20APR2025	Bangalore	45	0	800000	VALID	VALID	18MAR2025	20APR2025	KA01-D1004-Bangalore
5	D1005	Mohan	Radha	AP02	lawyer5@court.in	Joint	Active			Vijayawada	8	3	.	INVALID	VALID	.	.	AP02-D1005-Vijayawada
6	D1006	Akash	Nisha	MH01	lawyer6@court.in	Mother	Closed	12APR2025	15MAY2025	Mumbai	.	1	900000	VALID	VALID	12APR2025	15MAY2025	MH01-D1006-Mumbai
7	D1007	Arjun	Priya	XX01	lawyer7@court.in	Joint	Closed	10-10-2025	12-12-2025	Delhi	15	2	300000	VALID	VALID	10OCT2025	12DEC2025	XX01-D1007-Delhi
8	D1008	Tarun	Pooja	AP03	lawyer8@court.in	Father	Active	20JAN2025		Hyderabad	7	.	450000	VALID	VALID	20JAN2025	.	AP03-D1008-Hyderabad
9	D1009	Raj	Divya	KA02	lawyer9@court.in	Joint	Closed	15FEB2025	18MAR2025	Mysore	14	2	1000000	VALID	VALID	15FEB2025	18MAR2025	KA02-D1009-Mysore
10	D1010	Naveen	Sneha	AP04	lawyer10@court.in	Mother	Active	17MAR2025		Vizag	5	1	550000	VALID	VALID	17MAR2025	.	AP04-D1010-Vizag
11	D1011	Rakesh	Simran	MH02	lawyer11@court.in	Joint	Closed	18APR2025	22JUN2025	Pune	20	2	850000	VALID	VALID	18APR2025	22JUN2025	MH02-D1011-Pune
12	D1012	Ajay	Priti	TN02	lawyer12@court.in	Father	Closed	10MAY2025	20JUL2025	Chennai	18	1	620000	VALID	VALID	10MAY2025	20JUL2025	TN02-D1012-Chennai
13	D1013	Deepak	Rani	AP05	lawyer13@court.in	Joint	Active	14JUN2025		Guntur	13	2	580000	VALID	VALID	14JUN2025	.	AP05-D1013-Guntur
14	D1014	Vinod	Rekha	KA03	lawyer14@court.in	Mother	Closed	15JUL2025	18SEP2025	Hubli	9	0	400000	VALID	VALID	15JUL2025	18SEP2025	KA03-D1014-Hubli
15	D1015	Rahul	Keerthi	AP06	lawyer15@court.in	Joint	Active	21AUG2025		Tirupati	16	2	760000	VALID	VALID	21AUG2025	.	AP06-D1015-Tirupati
16	D1016	Mahesh	Swathi	TN03	lawyer16@court.in	Father	Closed	12SEP2025	15NOV2025	Madurai	4	1	250000	VALID	VALID	12SEP2025	15NOV2025	TN03-D1016-Madurai
17	D1017	Dinesh	Latha	MH03	lawyer17@court.in	Joint	Active	17OCT2025		Nagpur	11	3	690000	VALID	VALID	17OCT2025	.	MH03-D1017-Nagpur
18	D1018	Harish	Deepa	AP07	lawyer18@court.in	Mother	Closed	20NOV2025	22DEC2025	Nellore	6	0	350000	VALID	VALID	20NOV2025	22DEC2025	AP07-D1018-Nellore
19	D1019	Vijay	Sita	KA04	lawyer19@court.in	Joint	Closed	05DEC2025	10JAN2026	Belgaum	21	2	920000	VALID	VALID	05DEC2025	10JAN2026	KA04-D1019-Belgaum
20	D1020	Anand	Jyothi	AP08	lawyer20@court.in	Father	Active	12DEC2025		Kurnool	10	1	480000	VALID	VALID	12DEC2025	.	AP08-D1020-Kurnool

Why These Functions Matter

VERIFY / NOTDIGIT

Used to identify unwanted characters.

notdigit("abc500")

Returns position of first non-numeric character.

Very useful for:

Salary fields
Settlement values
Account numbers

CATX

Creates meaningful IDs.

Case_Label=

catx('-',Court_Code,Case_ID,City);

Output:

AP01-D1001-Hyderabad

Useful for:

Tracking
Auditing
Reporting

INPUT

Converts dates dynamically.

inputn(Filing_Date_Raw,'ANYDTDTE20.')

Handles:

12-01-2025

12JAN2025

2025-01-12

without writing separate conversion code.

Explanation:

This SAS DATA step is designed to transform the raw divorce dataset into a clean, standardized, and analysis-ready dataset. The statement data divorce_clean; set divorce_raw; creates a new dataset called DIVORCE_CLEAN by reading observations from the raw dataset DIVORCE_RAW. The first set of transformations focuses on standardizing text variables. lowcase(strip(Lawyer_Email)) removes unnecessary leading and trailing spaces from the lawyer email field and converts all email addresses to lowercase, ensuring consistency when validating or reporting email information. upcase(Court_Code) converts court codes to uppercase so that values such as "ap01", "Ap01", and "AP01" are treated identically. Similarly, propcase(City) converts city names into proper case format, improving readability in reports and dashboards.

The next section performs business-rule validation on the Years_Married variable. Divorce records with negative marriage durations are logically impossible, so values less than zero are converted to missing (.). Likewise, values greater than sixty years are considered unrealistic for this project and are also converted to missing. This prevents extreme or invalid values from distorting summary statistics and analytical results.

The settlement amount cleaning logic converts the character variable Settlement_Amount into a numeric variable called Settlement_Num. The input(Settlement_Amount,?? best12.) function attempts to convert character values into numbers while suppressing invalid data conversion notes through the ?? modifier. For example, values such as "500000" and "900000" are successfully converted, whereas invalid values like "abc500" become missing. The abs() function then converts negative settlement amounts into positive values, ensuring standardized monetary reporting. A new variable called Settlement_Flag is created to indicate whether the settlement amount is valid or invalid. Records with successfully converted amounts receive a value of "VALID", while records with missing converted values receive "INVALID".

The program then validates email addresses using the find() function. If the @ symbol is absent from the email address, the record is flagged as "INVALID" in the Email_Flag variable; otherwise, it is marked "VALID". This provides a simple but effective quality check for email data. Next, the code converts the raw filing and final order date fields into true SAS date values using the ANYDTDTE. informat. This informat is particularly useful because it can interpret multiple date formats such as 12-01-2025, 15FEB2025, and 2025-03-15. After conversion, the FORMAT statement displays these dates using the DATE9. format, making them easier to read and interpret.

Finally, the CATX() function creates a new variable called Case_Label by combining the court code, case identifier, and city into a single descriptive value separated by hyphens. For example, a record may produce a label such as AP01-D1001-Hyderabad. This composite identifier can be useful for reporting, auditing, and tracking records across different systems. Overall, this DATA step demonstrates several important data-cleaning concepts, including standardization, validation, data type conversion, error handling, date processing, quality flag creation, and generation of business-friendly identifiers, all of which are essential skills for SAS programmers working with real-world datasets.

3.PROC FREQ Validation

proc freq data=divorce_clean;

tables Court_Code Custody_Status Case_Status

Email_Flag;

run;

OUTPUT:

The FREQ Procedure


Court_Code	Frequency	Percent	Cumulative Frequency	Cumulative Percent
AP01	2	10.00	2	10.00
AP02	1	5.00	3	15.00
AP03	1	5.00	4	20.00
AP04	1	5.00	5	25.00
AP05	1	5.00	6	30.00
AP06	1	5.00	7	35.00
AP07	1	5.00	8	40.00
AP08	1	5.00	9	45.00
KA01	1	5.00	10	50.00
KA02	1	5.00	11	55.00
KA03	1	5.00	12	60.00
KA04	1	5.00	13	65.00
MH01	1	5.00	14	70.00
MH02	1	5.00	15	75.00
MH03	1	5.00	16	80.00
TN01	1	5.00	17	85.00
TN02	1	5.00	18	90.00
TN03	1	5.00	19	95.00
XX01	1	5.00	20	100.00


Custody_Status	Frequency	Percent	Cumulative Frequency	Cumulative Percent
Father	5	25.00	5	25.00
Joint	10	50.00	15	75.00
Mother	5	25.00	20	100.00


Case_Status	Frequency	Percent	Cumulative Frequency	Cumulative Percent
Active	8	40.00	8	40.00
Closed	12	60.00	20	100.00


Email_Flag	Frequency	Percent	Cumulative Frequency	Cumulative Percent
INVALID	1	5.00	1	5.00
VALID	19	95.00	20	100.00

Purpose:

Find unexpected categories
Detect spelling mistakes
Validate business rules

4.PROC SUMMARY

proc summary data=divorce_clean;

class Court_Code;

var Settlement_Amount;

output out=court_summary mean= median= max= min=;

run;

proc print data = court_summary;

run;

OUTPUT:


Obs	Court_Code	_TYPE_	_FREQ_	Settlement_Amount
1		0	20	610526.32
2	AP01	1	2	500000.00
3	AP02	1	1	.
4	AP03	1	1	450000.00
5	AP04	1	1	550000.00
6	AP05	1	1	580000.00
7	AP06	1	1	760000.00
8	AP07	1	1	350000.00
9	AP08	1	1	480000.00
10	KA01	1	1	800000.00
11	KA02	1	1	1000000.00
12	KA03	1	1	400000.00
13	KA04	1	1	920000.00
14	MH01	1	1	900000.00
15	MH02	1	1	850000.00
16	MH03	1	1	690000.00
17	TN01	1	1	700000.00
18	TN02	1	1	620000.00
19	TN03	1	1	250000.00
20	XX01	1	1	300000.00

Purpose:

Court-wise settlement analysis.

5.PROC RANK

proc rank data=divorce_clean out=ranked_cases groups=4;

var Settlement_Amount;

ranks Settlement_Group;

run;

proc print data = ranked_cases;

run;

OUTPUT:


Obs	Case_ID	Husband_Name	Wife_Name	Court_Code	Lawyer_Email	Custody_Status	Case_Status	Filing_Date_Raw	Final_Order_Raw	City	Years_Married	Children_Count	Settlement_Amount	Settlement_Flag	Email_Flag	Filing_Date	Final_Order	Case_Label	Settlement_Group
1	D1001	Ravi	Anjali	AP01	lawyer1@court.in	Joint	Closed	12-01-2025	15-04-2025	Hyderabad	12	2	500000	VALID	VALID	01DEC2025	15APR2025	AP01-D1001-Hyderabad	1
2	D1002	Ravi	Anjali	AP01	lawyer1@court.in	Joint	Closed	12-01-2025	15-04-2025	Hyderabad	12	2	500000	VALID	VALID	01DEC2025	15APR2025	AP01-D1002-Hyderabad	1
3	D1003	Kiran	Meena	TN01	invalidmail	Mother	Active	15/02/2025		Chennai	.	1	700000	VALID	INVALID	15FEB2025	.	TN01-D1003-Chennai	2
4	D1004	Suresh	Kavya	KA01	lawyer4@court.in	Father	Closed	18MAR2025	20APR2025	Bangalore	45	0	800000	VALID	VALID	18MAR2025	20APR2025	KA01-D1004-Bangalore	3
5	D1005	Mohan	Radha	AP02	lawyer5@court.in	Joint	Active			Vijayawada	8	3	.	INVALID	VALID	.	.	AP02-D1005-Vijayawada	.
6	D1006	Akash	Nisha	MH01	lawyer6@court.in	Mother	Closed	12APR2025	15MAY2025	Mumbai	.	1	900000	VALID	VALID	12APR2025	15MAY2025	MH01-D1006-Mumbai	3
7	D1007	Arjun	Priya	XX01	lawyer7@court.in	Joint	Closed	10-10-2025	12-12-2025	Delhi	15	2	300000	VALID	VALID	10OCT2025	12DEC2025	XX01-D1007-Delhi	0
8	D1008	Tarun	Pooja	AP03	lawyer8@court.in	Father	Active	20JAN2025		Hyderabad	7	.	450000	VALID	VALID	20JAN2025	.	AP03-D1008-Hyderabad	1
9	D1009	Raj	Divya	KA02	lawyer9@court.in	Joint	Closed	15FEB2025	18MAR2025	Mysore	14	2	1000000	VALID	VALID	15FEB2025	18MAR2025	KA02-D1009-Mysore	3
10	D1010	Naveen	Sneha	AP04	lawyer10@court.in	Mother	Active	17MAR2025		Vizag	5	1	550000	VALID	VALID	17MAR2025	.	AP04-D1010-Vizag	1
11	D1011	Rakesh	Simran	MH02	lawyer11@court.in	Joint	Closed	18APR2025	22JUN2025	Pune	20	2	850000	VALID	VALID	18APR2025	22JUN2025	MH02-D1011-Pune	3
12	D1012	Ajay	Priti	TN02	lawyer12@court.in	Father	Closed	10MAY2025	20JUL2025	Chennai	18	1	620000	VALID	VALID	10MAY2025	20JUL2025	TN02-D1012-Chennai	2
13	D1013	Deepak	Rani	AP05	lawyer13@court.in	Joint	Active	14JUN2025		Guntur	13	2	580000	VALID	VALID	14JUN2025	.	AP05-D1013-Guntur	2
14	D1014	Vinod	Rekha	KA03	lawyer14@court.in	Mother	Closed	15JUL2025	18SEP2025	Hubli	9	0	400000	VALID	VALID	15JUL2025	18SEP2025	KA03-D1014-Hubli	0
15	D1015	Rahul	Keerthi	AP06	lawyer15@court.in	Joint	Active	21AUG2025		Tirupati	16	2	760000	VALID	VALID	21AUG2025	.	AP06-D1015-Tirupati	2
16	D1016	Mahesh	Swathi	TN03	lawyer16@court.in	Father	Closed	12SEP2025	15NOV2025	Madurai	4	1	250000	VALID	VALID	12SEP2025	15NOV2025	TN03-D1016-Madurai	0
17	D1017	Dinesh	Latha	MH03	lawyer17@court.in	Joint	Active	17OCT2025		Nagpur	11	3	690000	VALID	VALID	17OCT2025	.	MH03-D1017-Nagpur	2
18	D1018	Harish	Deepa	AP07	lawyer18@court.in	Mother	Closed	20NOV2025	22DEC2025	Nellore	6	0	350000	VALID	VALID	20NOV2025	22DEC2025	AP07-D1018-Nellore	0
19	D1019	Vijay	Sita	KA04	lawyer19@court.in	Joint	Closed	05DEC2025	10JAN2026	Belgaum	21	2	920000	VALID	VALID	05DEC2025	10JAN2026	KA04-D1019-Belgaum	3
20	D1020	Anand	Jyothi	AP08	lawyer20@court.in	Father	Active	12DEC2025		Kurnool	10	1	480000	VALID	VALID	12DEC2025	.	AP08-D1020-Kurnool	1

Produces:

0 = Lowest

1 = Low

2 = Medium

3 = Highest

6.Create Raw Divorce Dataset in R

library(dplyr)

library(stringr)

library(lubridate)

library(readr)

divorce_raw <- data.frame(

case_id =

c("D1001","D1002","D1003","D1004",

"D1005","D1006","D1007","D1008",

"D1009","D1010","D1011","D1012"),

years_married =

c(12,12,-2,45,8,110,15,7,14,5,20,18),

settlement_amount =

c("500000","500000","700000",

"-800000","abc500",

"900000","300000",

"450000","1000000",

"550000","850000",

"620000"),

court_code =

c("AP01","AP01","TN01",

"KA01","AP02","MH01",

"XX01","AP03",

"KA02","AP04",

"MH02","TN02")

)

OUTPUT:

	case_id	years_married	settlement_amount	court_code
1	D1001	12	500000	AP01
2	D1002	12	500000	AP01
3	D1003	-2	700000	TN01
4	D1004	45	-800000	KA01
5	D1005	8	abc500	AP02
6	D1006	110	900000	MH01
7	D1007	15	300000	XX01
8	D1008	7	450000	AP03
9	D1009	14	1000000	KA02
10	D1010	5	550000	AP04
11	D1011	20	850000	MH02
12	D1012	18	620000	TN02

Instead of:

across()

str_trim()

str_to_title()

Use:

library(dplyr)

library(stringr)

library(lubridate)

library(readr)

7.R Cleaning Pipeline

divorce_clean <- divorce_raw %>%

mutate(

court_code = toupper(court_code),

years_married =

replace(years_married,years_married < 0 |

years_married > 60,NA),

settlement_amount =parse_number(

settlement_amount),

settlement_amount = abs(settlement_amount),

case_group =case_when(

settlement_amount >= 800000 ~ "HIGH",

settlement_amount >= 500000 ~ "MEDIUM",

TRUE ~ "LOW")

)

OUTPUT:

	case_id	years_married	settlement_amount	court_code	case_group
1	D1001	12	500000	AP01	MEDIUM
2	D1002	12	500000	AP01	MEDIUM
3	D1003	NA	700000	TN01	MEDIUM
4	D1004	45	800000	KA01	HIGH
5	D1005	8	500	AP02	LOW
6	D1006	NA	900000	MH01	HIGH
7	D1007	15	300000	XX01	LOW
8	D1008	7	450000	AP03	LOW
9	D1009	14	1000000	KA02	HIGH
10	D1010	5	550000	AP04	MEDIUM
11	D1011	20	850000	MH02	HIGH
12	D1012	18	620000	TN02	MEDIUM

Why parse_number() Is Powerful

Instead of:

gsub("[^0-9]","",x)

Use:

parse_number()

Examples:

parse_number("$500000")

parse_number("abc500")

parse_number("-800000")

Automatically extracts numbers.

Advanced Divorce Analytics in R

8.Court-wise Statistics

divorce_group <- divorce_clean %>%

group_by(court_code) %>%

summarise(cases=n(),

avg_settlement=mean(

settlement_amount,na.rm=TRUE)

)

OUTPUT:

	court_code	cases	avg_settlement
1	AP01	2	500000
2	AP02	1	500
3	AP03	1	450000
4	AP04	1	550000
5	KA01	1	800000
6	KA02	1	1000000
7	MH01	1	900000
8	MH02	1	850000
9	TN01	1	700000
10	TN02	1	620000
11	XX01	1	300000

9.Detect Duplicates

# Duplicate Check

divorce_summary <- divorce_clean %>%

count(case_id) %>%

filter(n > 1)

# Missing Value Audit

colSums(is.na(divorce_clean))

CONSOLE:

> colSums(is.na(divorce_clean))

case_id years_married settlement_amount court_code

0 2 0 0

case_group

# Structure Check

str(divorce_clean)

CONSOLE:

> str(divorce_clean)

'data.frame': 12 obs. of 5 variables:

$ case_id : chr "D1001" "D1002" "D1003" "D1004" ...

$ years_married : num 12 12 NA 45 8 NA 15 7 14 5 ...

$ settlement_amount: num 500000 500000 700000 800000 500 900000 300000 450000 1000000 550000 ...

$ court_code : chr "AP01" "AP01" "TN01" "KA01" ...

$ case_group : chr "MEDIUM" "MEDIUM" "MEDIUM" "HIGH" ...

# Summary Statistics

summary(divorce_clean)

CONSOLE:

summary(divorce_clean)

case_id years_married settlement_amount court_code

Length:12 Min. : 5.00 Min. : 500 Length:12

Class :character 1st Qu.: 9.00 1st Qu.: 487500 Class :character

Mode :character Median :13.00 Median : 585000 Mode :character

Mean :15.60 Mean : 597542

3rd Qu.:17.25 3rd Qu.: 812500

Max. :45.00 Max. :1000000

NA's :2

case_group

Length:12

Class :character

Mode :character

# Invalid Settlement Records

divorce_clean %>%

filter(is.na(settlement_amount))

CONSOLE:

[1] case_id years_married settlement_amount court_code

[5] case_group

<0 rows> (or 0-length row.names)

# Invalid Marriage Duration

divorce_clean %>%

filter(is.na(years_married))

CONSOLE:

case_id years_married settlement_amount court_code case_group

1 D1003 NA 7e+05 TN01 MEDIUM

2 D1006 NA 9e+05 MH01 HIGH

SAS vs R Comparison for Divorce Data

Activity	SAS	R
Date Conversion	INPUTN	lubridate
Categorization	FORMAT	case_when
Duplicate Detection	PROC SORT	count()
Validation	PROC FREQ	table()
Aggregation	PROC SUMMARY	summarise()
Ranking	PROC RANK	ntile()

Interview Questions & Answers

1. Data Integrity Investigation

Question:
A divorce court suddenly reports 40% more finalized cases. How would you validate whether the increase is real or caused by duplicate records?

Short Answer:
I would compare current and historical case counts, check for duplicate Case_ID values using PROC SORT NODUPKEY or PROC SQL, and verify whether additional records are genuine new cases or duplicated entries from data loading or merging issues.

2. Settlement Amount Corruption

Question:
A settlement column contains values like 500000, -700000, and abc500. How would you clean and validate these values in SAS?

Short Answer:
I would use the INPUT() function to convert character values to numeric, apply ABS() to handle negative amounts if required by business rules, and flag invalid values such as abc500 as missing for further investigation.

3. Court Code Governance

Question:
How would you identify invalid court jurisdictions such as XX01 before reporting?

Short Answer:
I would compare court codes against a predefined list of valid court codes using PROC FORMAT, PROC SQL, or lookup tables and flag any unmatched codes as invalid before generating reports.

4. Date Standardization Challenge

Question:
What approach would you use when filing dates arrive in five different formats?

Short Answer:
I would use the ANYDTDTE. informat to convert all date formats into a standard SAS date value and then apply a common display format such as DATE9. for consistency.

5. Enterprise Validation Strategy

Question:
How would you independently QC a divorce dataset generated by another programmer?

Short Answer:
I would independently recreate the dataset using the specifications, compare record counts, variables, summary statistics, and key outputs, then investigate and resolve any discrepancies found during reconciliation.

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

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. Here we learn about DIVORCE DATA.

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

This project is suitable for:
·  Students learning SAS
·  Data analysts building portfolios
·  Professionals preparing for SAS interviews
·  Bloggers writing about analytics and smart cities
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Follow Us On :

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

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

---> Support Us By Following Our Blog..

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

1.How Can SAS Map The Journey And Influence Of India’s Freedom Fighters?

2.Can SAS Identify The Best Project And Tool Combinations For Software Teams?

3.Can SAS Reveal Which Office Environments Are Truly Productive And Which Are Not?

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

About Us | Contact | Privacy Policy

Search This Blog

SAS Learning Hub