381.Can SAS Uncover Call Drops, Network Congestion, and Fraud from Telecom Call Records?

Can SAS Uncover Call Drops, Network Congestion, and Fraud from Telecom Call Records?

HERE IN THIS PROJECT WE USED THESE SAS STATEMENTS —
DATA STEP | PROC SQL |  PROC PRINT | PROC SGPLOT | MACROS | PROC CORR | PROC MEANS | PROC FREQ | PROC UNIVARIATE | PROC DATASETS DELETE

1. Introduction – Why Telecom Call Data Analytics Matters

Every second, millions of phone calls, video calls, roaming sessions, and data connections are created across telecom networks worldwide. Each of these interactions generates a Call Detail Record (CDR) containing critical information such as caller identity, call duration, network load, drop rate, and billing amount.

Telecom companies like Airtel, Jio, Vodafone, AT&T, and Verizon rely heavily on these records to monitor:

·       Network congestion and capacity

·       Call quality and drop rates

·       Revenue generation and leakage

·       Fraudulent calling behavior

·       Regional performance and infrastructure health

With growing 5G adoption and increasing digital traffic, analyzing call records has become a mission-critical business function. Even a small percentage of undetected call drops or fraud can lead to massive financial losses and customer dissatisfaction.

This project simulates a real-world telecom analytics environment using SAS, where call records are processed, cleaned, analyzed, and converted into business-ready insights.

2. What This Telecom Analytics Project Simulates

This SAS project replicates the complete data pipeline used inside telecom companies. It is designed to behave like a real operational system where:

·       Millions of daily calls are recorded

·       Network congestion is monitored

·       Revenue is calculated per region

·       Fraud is detected using business rules

·       Performance is summarized for management

The dataset includes realistic telecom variables such as:

·       Caller ID – Identifies each subscriber

·       Region – Geographic location of the call

·       Call Duration – Network usage in minutes

·       Call Type – Voice, Data, Roaming, or International

·       Network Load – Network congestion level

·       Drop Rate – Quality of service indicator

·       Revenue – Money earned from the call

·       Call Date & Billing Month – Time-based analysis

Using these variables, the project demonstrates how telecom operators make data-driven decisions to optimize performance and protect revenue.

3. How This Project Reflects Real Industry Use Cases

The logic implemented in this project closely resembles what is used in:

·       Telecom billing systems

·       Fraud monitoring platforms

·       Network operations centers

·       Regulatory reporting systems

·       Customer experience analytics

4. Why SAS Is Used for Telecom Analytics

SAS is widely used in the telecom industry because it provides:

·       High-performance data processing

·       Powerful SQL-based analytics

·       Advanced statistical procedures

·       Strong reporting and visualization

·       Reliable automation through macros

Telecom companies trust SAS for handling billions of call records with accuracy, speed, and auditability. This project demonstrates how SAS can be used to build a complete telecom analytics solution from raw data to executive-ready insights.

5. Who Should Study This Project

This tutorial is ideal for:

·       SAS learners and programmers

·       Data analysts and data scientists

·       Telecom analytics professionals

·       Interview preparation candidates

·       Blog readers learning real-world analytics

6. BUSINESS CONTEXT

Telecom operators manage millions of calls per day.
Each call generates a Call Detail Record (CDR) with:

Field	Purpose
Caller ID	Who made the call
Region	Where call originated
Call Duration	Network usage
Call Type	Voice / Data / Roaming / International
Network Load	Congestion
Drop Rate	Call failures
Revenue	Billing impact
Call Date	Time intelligence

Telecom companies must:

• Control network congestion
• Detect revenue leakage & fraud
• Improve customer experience
• Optimize tower & spectrum usage

7. SAS PROGRAM

STEP-1 — Create Telecom Call Dataset

data telecom_raw;

    length Caller_ID $6 Region $10 Call_Type $15;

    format Call_Date date9.;

    input Caller_ID $ Region $ Call_Date : date9. Call_Duration Network_Load Drop_Rate 

          Revenue Call_Type;

    datalines;

C001 South 01JAN2025 15 78 2.5 120 Voice

C002 North 02JAN2025 22 82 3.1 210 Data

C003 East 03JAN2025 5 45 0.9 50 Voice

C004 West 04JAN2025 35 95 6.5 350 International

C005 South 05JAN2025 18 70 1.8 160 Voice

C006 North 06JAN2025 60 98 8.2 600 Roaming

C007 East 07JAN2025 12 55 1.2 100 Data

C008 West 08JAN2025 42 90 5.4 420 International

C009 South 09JAN2025 8 40 0.7 60 Voice

C010 North 10JAN2025 55 92 7.8 560 Roaming

C011 East 11JAN2025 25 75 3.5 230 Data

C012 West 12JAN2025 48 88 4.9 480 International

C013 South 13JAN2025 10 50 1.0 80 Voice

C014 North 14JAN2025 65 97 8.9 650 Roaming

C015 East 15JAN2025 20 72 2.2 190 Data

C016 West 16JAN2025 38 85 4.5 390 International

;

run;

proc print data=telecom_raw;

run;

OUTPUT:

Obs	Caller_ID	Region	Call_Type	Call_Date	Call_Duration	Network_Load	Drop_Rate	Revenue
1	C001	South	Voice	01JAN2025	15	78	2.5	120
2	C002	North	Data	02JAN2025	22	82	3.1	210
3	C003	East	Voice	03JAN2025	5	45	0.9	50
4	C004	West	International	04JAN2025	35	95	6.5	350
5	C005	South	Voice	05JAN2025	18	70	1.8	160
6	C006	North	Roaming	06JAN2025	60	98	8.2	600
7	C007	East	Data	07JAN2025	12	55	1.2	100
8	C008	West	International	08JAN2025	42	90	5.4	420
9	C009	South	Voice	09JAN2025	8	40	0.7	60
10	C010	North	Roaming	10JAN2025	55	92	7.8	560
11	C011	East	Data	11JAN2025	25	75	3.5	230
12	C012	West	International	12JAN2025	48	88	4.9	480
13	C013	South	Voice	13JAN2025	10	50	1.0	80
14	C014	North	Roaming	14JAN2025	65	97	8.9	650
15	C015	East	Data	15JAN2025	20	72	2.2	190
16	C016	West	International	16JAN2025	38	85	4.5	390

• Simulates telecom call records
• Each row = one call
• Date handled via DATE9
• Call type classification
• Revenue tracking

STEP-2 — Clean & Standardize Text

data telecom_clean;

    set telecom_raw;

    Caller_ID = strip(upcase(Caller_ID));

    Region = propcase(trim(Region));

    Call_Type = propcase(Call_Type);

run;

proc print data=telecom_clean;

run;

OUTPUT:

Obs	Caller_ID	Region	Call_Type	Call_Date	Call_Duration	Network_Load	Drop_Rate	Revenue
1	C001	South	Voice	01JAN2025	15	78	2.5	120
2	C002	North	Data	02JAN2025	22	82	3.1	210
3	C003	East	Voice	03JAN2025	5	45	0.9	50
4	C004	West	International	04JAN2025	35	95	6.5	350
5	C005	South	Voice	05JAN2025	18	70	1.8	160
6	C006	North	Roaming	06JAN2025	60	98	8.2	600
7	C007	East	Data	07JAN2025	12	55	1.2	100
8	C008	West	International	08JAN2025	42	90	5.4	420
9	C009	South	Voice	09JAN2025	8	40	0.7	60
10	C010	North	Roaming	10JAN2025	55	92	7.8	560
11	C011	East	Data	11JAN2025	25	75	3.5	230
12	C012	West	International	12JAN2025	48	88	4.9	480
13	C013	South	Voice	13JAN2025	10	50	1.0	80
14	C014	North	Roaming	14JAN2025	65	97	8.9	650
15	C015	East	Data	15JAN2025	20	72	2.2	190
16	C016	West	International	16JAN2025	38	85	4.5	390

• Removes spaces
• Standardizes case
• Improves joins
• Avoids duplicates
• Production readiness

STEP-3 — Add Billing Month Using INTNX

data telecom_dates;

    set telecom_clean;

    Bill_Month = intnx('month', Call_Date, 0, 'b');

    format Bill_Month monyy7.;

run;

proc print data=telecom_dates;

run;

OUTPUT:

Obs	Caller_ID	Region	Call_Type	Call_Date	Call_Duration	Network_Load	Drop_Rate	Revenue	Bill_Month
1	C001	South	Voice	01JAN2025	15	78	2.5	120	JAN2025
2	C002	North	Data	02JAN2025	22	82	3.1	210	JAN2025
3	C003	East	Voice	03JAN2025	5	45	0.9	50	JAN2025
4	C004	West	International	04JAN2025	35	95	6.5	350	JAN2025
5	C005	South	Voice	05JAN2025	18	70	1.8	160	JAN2025
6	C006	North	Roaming	06JAN2025	60	98	8.2	600	JAN2025
7	C007	East	Data	07JAN2025	12	55	1.2	100	JAN2025
8	C008	West	International	08JAN2025	42	90	5.4	420	JAN2025
9	C009	South	Voice	09JAN2025	8	40	0.7	60	JAN2025
10	C010	North	Roaming	10JAN2025	55	92	7.8	560	JAN2025
11	C011	East	Data	11JAN2025	25	75	3.5	230	JAN2025
12	C012	West	International	12JAN2025	48	88	4.9	480	JAN2025
13	C013	South	Voice	13JAN2025	10	50	1.0	80	JAN2025
14	C014	North	Roaming	14JAN2025	65	97	8.9	650	JAN2025
15	C015	East	Data	15JAN2025	20	72	2.2	190	JAN2025
16	C016	West	International	16JAN2025	38	85	4.5	390	JAN2025

• Groups by month
• Enables trends
• Supports billing
• Works with PROC SQL
• Date normalization

STEP-4 — Network Utilization Macro

%macro utilization(input=, output=);

data &output;

    set &input;

    length Utilization $10;

    if Network_Load < 60 then Utilization="Low";

    else if Network_Load < 85 then Utilization="Medium";

    else Utilization="High";

run;

proc print data=&output;

run;

%mend;

%utilization(input=telecom_dates, output=telecom_util);

OUTPUT:

Obs	Caller_ID	Region	Call_Type	Call_Date	Call_Duration	Network_Load	Drop_Rate	Revenue	Bill_Month	Utilization
1	C001	South	Voice	01JAN2025	15	78	2.5	120	JAN2025	Medium
2	C002	North	Data	02JAN2025	22	82	3.1	210	JAN2025	Medium
3	C003	East	Voice	03JAN2025	5	45	0.9	50	JAN2025	Low
4	C004	West	International	04JAN2025	35	95	6.5	350	JAN2025	High
5	C005	South	Voice	05JAN2025	18	70	1.8	160	JAN2025	Medium
6	C006	North	Roaming	06JAN2025	60	98	8.2	600	JAN2025	High
7	C007	East	Data	07JAN2025	12	55	1.2	100	JAN2025	Low
8	C008	West	International	08JAN2025	42	90	5.4	420	JAN2025	High
9	C009	South	Voice	09JAN2025	8	40	0.7	60	JAN2025	Low
10	C010	North	Roaming	10JAN2025	55	92	7.8	560	JAN2025	High
11	C011	East	Data	11JAN2025	25	75	3.5	230	JAN2025	Medium
12	C012	West	International	12JAN2025	48	88	4.9	480	JAN2025	High
13	C013	South	Voice	13JAN2025	10	50	1.0	80	JAN2025	Low
14	C014	North	Roaming	14JAN2025	65	97	8.9	650	JAN2025	High
15	C015	East	Data	15JAN2025	20	72	2.2	190	JAN2025	Medium
16	C016	West	International	16JAN2025	38	85	4.5	390	JAN2025	High

• Categorizes traffic
• Reusable logic
• Business rules
• Automation
• Network optimization

STEP-5 — Fraud Risk Macro

%macro fraud(input=, output=);

data &output;

    set &input;

    length Fraud_Flag $5;

    if Call_Duration > 50 and Revenue > 500 then Fraud_Flag="YES";

    else Fraud_Flag="NO";

run;

proc print data=&output;

run;

%mend;

%fraud(input=telecom_util, output=telecom_fraud);

OUTPUT:

Obs	Caller_ID	Region	Call_Type	Call_Date	Call_Duration	Network_Load	Drop_Rate	Revenue	Bill_Month	Utilization	Fraud_Flag
1	C001	South	Voice	01JAN2025	15	78	2.5	120	JAN2025	Medium	NO
2	C002	North	Data	02JAN2025	22	82	3.1	210	JAN2025	Medium	NO
3	C003	East	Voice	03JAN2025	5	45	0.9	50	JAN2025	Low	NO
4	C004	West	International	04JAN2025	35	95	6.5	350	JAN2025	High	NO
5	C005	South	Voice	05JAN2025	18	70	1.8	160	JAN2025	Medium	NO
6	C006	North	Roaming	06JAN2025	60	98	8.2	600	JAN2025	High	YES
7	C007	East	Data	07JAN2025	12	55	1.2	100	JAN2025	Low	NO
8	C008	West	International	08JAN2025	42	90	5.4	420	JAN2025	High	NO
9	C009	South	Voice	09JAN2025	8	40	0.7	60	JAN2025	Low	NO
10	C010	North	Roaming	10JAN2025	55	92	7.8	560	JAN2025	High	YES
11	C011	East	Data	11JAN2025	25	75	3.5	230	JAN2025	Medium	NO
12	C012	West	International	12JAN2025	48	88	4.9	480	JAN2025	High	NO
13	C013	South	Voice	13JAN2025	10	50	1.0	80	JAN2025	Low	NO
14	C014	North	Roaming	14JAN2025	65	97	8.9	650	JAN2025	High	YES
15	C015	East	Data	15JAN2025	20	72	2.2	190	JAN2025	Medium	NO
16	C016	West	International	16JAN2025	38	85	4.5	390	JAN2025	High	NO

• Finds abnormal calls
• Uses duration + revenue
• Flags risky records
• Telecom risk model
• Audit support

STEP-6 — PROC SQL Revenue Summary

proc sql;

create table revenue_summary as

select Region,

       sum(Revenue) as Total_Revenue,

       avg(Drop_Rate) as Avg_Drop

from telecom_fraud

group by Region;

quit;

proc print data=revenue_summary;

run;

OUTPUT:

Obs	Region	Total_Revenue	Avg_Drop
1	East	570	1.950
2	North	2020	7.000
3	South	420	1.500
4	West	1640	5.325

• Summarizes data
• Region comparison
• Business reporting
• KPI generation
• SQL analytics

STEP-7 — PROC FREQ

proc freq data=telecom_fraud;

 tables Fraud_Flag*Call_Type;

run;

OUTPUT:

The FREQ Procedure

Frequency Percent Row Pct Col Pct

Table of Fraud_Flag by Call_Type Fraud_Flag Call_Type Data International Roaming Voice Total NO 4 25.00 30.77 100.00 4 25.00 30.77 100.00 0 0.00 0.00 0.00 5 31.25 38.46 100.00 13 81.25     YES 0 0.00 0.00 0.00 0 0.00 0.00 0.00 3 18.75 100.00 100.00 0 0.00 0.00 0.00 3 18.75     Total 4 25.00 4 25.00 3 18.75 5 31.25 16 100.00

• Cross analysis
• Fraud by type
• Monitoring tool
• Decision support
• Risk insight

STEP-8 — PROC MEANS

proc means data=telecom_fraud mean max min;

 var Call_Duration Network_Load Revenue Drop_Rate;

run;

OUTPUT:

The MEANS Procedure

Variable	Mean	Maximum	Minimum
Call_Duration Network_Load Revenue Drop_Rate	29.8750000 75.7500000 290.6250000 3.9437500	65.0000000 98.0000000 650.0000000 8.9000000	5.0000000 40.0000000 50.0000000 0.7000000

• Central tendency
• Extremes
• Network quality
• Revenue spread
• Operations control

STEP-9 — PROC UNIVARIATE

proc univariate data=telecom_fraud;

 var Revenue Call_Duration;

run;

OUTPUT:

The UNIVARIATE Procedure

Variable: Revenue

Moments
N	16	Sum Weights	16
Mean	290.625	Sum Observations	4650
Std Deviation	203.583521	Variance	41446.25
Skewness	0.49019766	Kurtosis	-1.1828857
Uncorrected SS	1973100	Corrected SS	621693.75
Coeff Variation	70.0502438	Std Error Mean	50.8958802

Basic Statistical Measures
Location		Variability
Mean	290.6250	Std Deviation	203.58352
Median	220.0000	Variance	41446
Mode	.	Range	600.00000
		Interquartile Range	340.00000

Tests for Location: Mu0=0
Test	Statistic		p Value
Student's t	t	5.710187	Pr > |t|	<.0001
Sign	M	8	Pr >= |M|	<.0001
Signed Rank	S	68	Pr >= |S|	<.0001

Quantiles (Definition 5)
Level	Quantile
100% Max	650
99%	650
95%	650
90%	600
75% Q3	450
50% Median	220
25% Q1	110
10%	60
5%	50
1%	50
0% Min	50

Extreme Observations
Lowest		Highest
Value	Obs	Value	Obs
50	3	420	8
60	9	480	12
80	13	560	10
100	7	600	6
120	1	650	14

The UNIVARIATE Procedure

Variable: Call_Duration

Moments
N	16	Sum Weights	16
Mean	29.875	Sum Observations	478
Std Deviation	19.4897409	Variance	379.85
Skewness	0.5041715	Kurtosis	-1.0629722
Uncorrected SS	19978	Corrected SS	5697.75
Coeff Variation	65.2376264	Std Error Mean	4.87243522

Basic Statistical Measures
Location		Variability
Mean	29.87500	Std Deviation	19.48974
Median	23.50000	Variance	379.85000
Mode	.	Range	60.00000
		Interquartile Range	31.50000

Tests for Location: Mu0=0
Test	Statistic		p Value
Student's t	t	6.131431	Pr > |t|	<.0001
Sign	M	8	Pr >= |M|	<.0001
Signed Rank	S	68	Pr >= |S|	<.0001

Quantiles (Definition 5)
Level	Quantile
100% Max	65.0
99%	65.0
95%	65.0
90%	60.0
75% Q3	45.0
50% Median	23.5
25% Q1	13.5
10%	8.0
5%	5.0
1%	5.0
0% Min	5.0

Extreme Observations
Lowest		Highest
Value	Obs	Value	Obs
5	3	42	8
8	9	48	12
10	13	55	10
12	7	60	6
15	1	65	14

• Identifies anomalies
• Detects skewness
• Risk analysis
• Revenue leakage
• Performance monitoring

STEP-10 — PROC CORR

proc corr data=telecom_fraud;

 var Call_Duration Network_Load Revenue Drop_Rate;

run;

OUTPUT:

The CORR Procedure

4 Variables:	Call_Duration Network_Load Revenue Drop_Rate

Simple Statistics
Variable	N	Mean	Std Dev	Sum	Minimum	Maximum
Call_Duration	16	29.87500	19.48974	478.00000	5.00000	65.00000
Network_Load	16	75.75000	19.02455	1212	40.00000	98.00000
Revenue	16	290.62500	203.58352	4650	50.00000	650.00000
Drop_Rate	16	3.94375	2.75729	63.10000	0.70000	8.90000

Pearson Correlation Coefficients, N = 16 Prob > |r| under H0: Rho=0
	Call_Duration	Network_Load	Revenue	Drop_Rate
Call_Duration	1.00000	0.87194 <.0001	0.99907 <.0001	0.96639 <.0001
Network_Load	0.87194 <.0001	1.00000	0.86809 <.0001	0.89278 <.0001
Revenue	0.99907 <.0001	0.86809 <.0001	1.00000	0.96597 <.0001
Drop_Rate	0.96639 <.0001	0.89278 <.0001	0.96597 <.0001	1.00000

• Dependency check
• Congestion impact
• Drop analysis
• Financial link
• Engineering decisions

STEP-11 — PROC SGPLOT

proc sgplot data=telecom_fraud;

 scatter x=Network_Load y=Drop_Rate;

run;

OUTPUT:

• Congestion vs drops
• Easy interpretation
• Executive dashboards
• Operations insight
• Quality tracking

STEP-12 — PROC TRANSPOSE

proc transpose data=revenue_summary out=rev_wide;

 by Region;

 id Region;

 var Total_Revenue;

run;

proc print data=rev_wide;

run;

OUTPUT:

Obs	Region	_NAME_	East	North	South	West
1	East	Total_Revenue	570	.	.	.
2	North	Total_Revenue	.	2020	.	.
3	South	Total_Revenue	.	.	420	.
4	West	Total_Revenue	.	.	.	1640

• Converts rows to columns
• BI tools ready
• Excel export
• Reporting layout
• Management view

STEP-13 — PROC DATASETS DELETE

proc datasets lib=work nolist;

 delete telecom_raw telecom_clean telecom_dates;

quit;

OUTPUT:

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

• Removes temp tables
• Optimizes workspace
• Professional coding
• Faster SAS
• Production hygiene

FINAL CONCLUSION

This Telecom Analytics project replicates real telecom operations systems where companies:

• Monitor call traffic
• Optimize network capacity
• Detect fraud
• Protect revenue
• Improve customer experience

INTERVIEW QUESTIONS FOR YOU

1.What is the Program Data Vector (PDV)?

2.What is the difference between LENGTH and FORMAT in SAS?

3.What is the use of the KEEP and DROP statements?

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 Telecom 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.How Can a Startup Build a Complete Data Analytics System Using Sas?

2.How Do We Explore Global Durga Temple Data Using Sas?

3.Can SAS Reveal Which Indian Sarees Are Most Popular, Profitable, and In-Demand Across States?

About Us | Contact | Privacy Policy