426.Can Advanced SAS Programming Detect And Fix Errors In Semiconductor Fabrication Plant Data?

Cracking Semiconductor Fabrication Data Errors Using Advanced SAS Programming

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

HERE IN THIS PROJECT WE USED THESE SAS STATEMENTS —DATA STEP | LENGTH | INPUT | DATALINES | FORMAT | SET | IF-THEN | MERGE | PROC SORT | PROC PRINT | PROC APPEND | PROC TRANSPOSE | PROC DATASETS | MACRO | NUMERIC AND CHARACTER FUNCTIONS 

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

Table of Contents

1. Introduction
2. Semiconductor Industry Business Context
3. Dataset Design Overview
4. Variables Description
5. Intentional Errors Introduced
6. SAS Environment Setup
7. Code 1 – Creating Raw Semiconductor Dataset
8. Code 2 – Using Character Functions to Clean Data
9. Code 3 – Numeric Calculations and Derived Metrics
10. Code 4 – Date Functions (MDY, INTCK, INTNX)
11. Code 5 – Utilization Classification
12. Code 6 – Fraud Detection Macro
13. Code 7 – Dataset Merge Example
14. Code 8 – Dataset Append Example
15. Code 9 – Dataset Transpose
16. Code 10 – PROC DATASETS Delete
17. Code 11 – Detecting Data Errors
18. Code 12 – Creating Final Cleaned Dataset
19. Final Dataset Output
20. Key Project Insights
21. Summary
22. Conclusion

1. Introduction

Semiconductor fabrication plants (commonly called fabs) manufacture microchips used in electronics, automobiles, smartphones, and satellites. These facilities generate massive amounts of process data such as wafer batches, defect densities, yield percentages, and machine downtime.

Data integrity is extremely important in semiconductor manufacturing because:

·  Small data errors can lead to millions of dollars in production losses

·  Incorrect yield calculations may hide equipment failures

·  Fraudulent data manipulation could hide fabrication defects

This project demonstrates how Advanced SAS Programming can be used to:

·  Create semiconductor plant datasets

·  Introduce intentional data errors

·  Detect fraud and inconsistencies

·  Correct those errors using SAS functions and procedures

·  Generate a clean analytical dataset

The project will also demonstrate:

·  Character functions

·  Numeric functions

·  Macros

·  Date functions

·  Dataset manipulation techniques

2. Semiconductor Industry Business Context

Semiconductor fabs operate with highly precise manufacturing processes. A typical fabrication facility tracks:

Metric	Meaning
Wafer Batch	Group of wafers processed together
Defect Density	Number of defects per wafer area
Yield Percentage	Percentage of functional chips
Process Node	Technology size (5nm, 7nm, etc)
Downtime Minutes	Machine stoppage time
Utilization	Machine usage efficiency

Companies such as:

·  Intel

·  TSMC

·  Samsung

·  GlobalFoundries

track these metrics continuously.

However real-world manufacturing datasets often contain:

·  Missing values

·  Incorrect date formats

·  Fraudulent yield values

·  Duplicate records

·  Character formatting errors

3. Dataset Design Overview

Variables

Variable	Type	Description
Fab_ID	Character	Fabrication plant identifier
Wafer_Batch	Character	Wafer batch ID
Defect_Density	Numeric	Defects per wafer
Yield_Percentage	Numeric	Production yield
Process_Node	Character	Fabrication technology
Downtime_Minutes	Numeric	Equipment downtime
Utilization	Numeric	Equipment utilization
Production_Fees	Numeric	Production cost
Production_Date	Character	Production date

Code 1 — Creating Raw Semiconductor Dataset (With Intentional Errors)

data semiconductor_fab_raw;

input Fab_ID $ Wafer_Batch $ Defect_Density Yield_Percentage

Process_Node $ Downtime_Minutes Utilization Production_Fees Production_Date $;

datalines;

fab01 WB1001 0.32 92 7nm 120 85 45000 02-15-2024

fab02 WB1002 . 95 5nm 90 88 47000 03-10-2024

fab03 WB1003 0.55 105 10nm 300 110 42000 04-05-2024

fab04 WB1004 0.40 89 7nm -50 80 41000 05-01-2024

fab05 WB1005 0.28 94 5nm 60 75 39000 06-18-2024

fab06 WB1006 0.90 60 14nm 500 40 35000 07-22-2024

fab07 WB1007 . 97 5nm 40 92 48000 08-11-2024

fab08 WB1008 0.25 96 3nm 20 95 50000 09-15-2024

fab09 WB1009 0.45 91 7nm 200 70 43000 10-20-2024

fab10 WB1010 0.60 88 10nm 350 65 42000 11-12-2024

;

run;

proc print data=semiconductor_fab_raw;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date
1	fab01	WB1001	0.32	92	7nm	120	85	45000	02-15-20
2	fab02	WB1002	.	95	5nm	90	88	47000	03-10-20
3	fab03	WB1003	0.55	105	10nm	300	110	42000	04-05-20
4	fab04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20
5	fab05	WB1005	0.28	94	5nm	60	75	39000	06-18-20
6	fab06	WB1006	0.90	60	14nm	500	40	35000	07-22-20
7	fab07	WB1007	.	97	5nm	40	92	48000	08-11-20
8	fab08	WB1008	0.25	96	3nm	20	95	50000	09-15-20
9	fab09	WB1009	0.45	91	7nm	200	70	43000	10-20-20
10	fab10	WB1010	0.60	88	10nm	350	65	42000	11-12-20

Explanation

This program creates the initial semiconductor dataset.

However it intentionally includes several data quality issues.

Intentional Errors Introduced

Error 1 — Missing Defect Density

.

Meaning missing value.

Error 2 — Yield Above 100%

105

Yield cannot exceed 100%.

Error 3 — Negative Downtime

-50

Downtime cannot be negative.

Error 4 — Utilization Above 100%

110

Machine utilization cannot exceed 100%.

Why Create Intentional Errors?

In real manufacturing datasets:

·  Data entry mistakes occur

·  Sensor malfunctions produce invalid values

·  Fraudulent adjustments may hide process failures

Detecting such issues is an important SAS programmer skill.

Code 2 — Character Data Cleaning

data fab_clean_char;

set semiconductor_fab_raw;

Fab_ID = upcase(strip(Fab_ID));

Process_Node = propcase(Process_Node);

Batch_Label = catx("_", Fab_ID, Wafer_Batch);

run;

proc print data=fab_clean_char;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010

Explanation

This step cleans and standardizes character variables.

Character Functions Used

STRIP

Removes leading and trailing spaces.

strip(Fab_ID)

UPCASE

Converts characters to uppercase.

Example:

fab01 → FAB01

PROPERCASE

Converts text into readable format.

7nm → 7Nm

CATX

Concatenates values with delimiter.

Example output:

FAB01_WB1001

Why Character Cleaning Matters

In large datasets inconsistent formatting causes:

·  Merge failures

·  Duplicate keys

·  Incorrect groupings

Cleaning character variables ensures consistency.

Code 3 — Numeric Calculations

data fab_numeric;

set fab_clean_char;

Effective_Yield = Yield_Percentage - Defect_Density*10;

Downtime_Hours = Downtime_Minutes/60;

run;

proc print data=fab_numeric;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003	99.5	5.00000
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333

Explanation

This step derives additional numeric metrics.

Numeric Calculations

Effective Yield

Yield_Percentage – Defect_Density × 10

Used to estimate yield impact from defects.

Downtime Hours

Converts minutes to hours.

Why Numeric Calculations Are Useful

Engineers analyze derived metrics to evaluate:

·  Production efficiency

·  Equipment reliability

·  Yield performance

Code 4 — Date Conversion Using MDY

data fab_dates;

set fab_numeric;

Prod_Date = input(Production_Date, mmddyy10.);

format Prod_Date date9.;

run;

proc print data=fab_dates;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours	Prod_Date
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000	15FEB2020
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000	10MAR2020
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003	99.5	5.00000	05APR2020
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333	01MAY2020
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000	18JUN2020
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333	22JUL2020
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667	11AUG2020
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333	15SEP2020
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333	20OCT2020
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333	12NOV2020

Explanation

The dataset currently stores dates as character values.

This step converts them to SAS date format.

Code 5 — Using INTCK

data fab_dates2;

set fab_dates;

Days_Since_Production = intck('day', Prod_Date, today());

run;

proc print data=fab_dates2;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours	Prod_Date	Days_Since_Production
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000	15FEB2020	2217
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000	10MAR2020	2193
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003	99.5	5.00000	05APR2020	2167
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333	01MAY2020	2141
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000	18JUN2020	2093
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333	22JUL2020	2059
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667	11AUG2020	2039
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333	15SEP2020	2004
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333	20OCT2020	1969
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333	12NOV2020	1946

Purpose

INTCK calculates time intervals between dates.

Example:

INTCK('DAY', StartDate, EndDate)

Used in:

·  equipment aging analysis

·  production monitoring

Code 6 — Using INTNX

data fab_dates3;

set fab_dates2;

Next_Service_Date = intnx('month', Prod_Date, 3);

format Next_Service_Date date9.;

run;

proc print data=fab_dates3;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours	Prod_Date	Days_Since_Production	Next_Service_Date
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000	15FEB2020	2217	01MAY2020
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000	10MAR2020	2193	01JUN2020
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003	99.5	5.00000	05APR2020	2167	01JUL2020
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333	01MAY2020	2141	01AUG2020
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000	18JUN2020	2093	01SEP2020
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333	22JUL2020	2059	01OCT2020
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667	11AUG2020	2039	01NOV2020
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333	15SEP2020	2004	01DEC2020
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333	20OCT2020	1969	01JAN2021
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333	12NOV2020	1946	01FEB2021

Purpose

INTNX calculates future or past dates.

Here it schedules maintenance every 3 months.

Code 7 — Utilization Classification

data fab_utilization;

set fab_dates3;

length Util_Class $20;

if Utilization >=90 then Util_Class="High Utilization";

else if Utilization >=70 then Util_Class="Moderate Utilization";

else Util_Class="Low Utilization";

run;

proc print data=fab_utilization;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours	Prod_Date	Days_Since_Production	Next_Service_Date	Util_Class
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000	15FEB2020	2217	01MAY2020	Moderate Utilization
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000	10MAR2020	2193	01JUN2020	Moderate Utilization
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003	99.5	5.00000	05APR2020	2167	01JUL2020	High Utilization
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333	01MAY2020	2141	01AUG2020	Moderate Utilization
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000	18JUN2020	2093	01SEP2020	Moderate Utilization
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333	22JUL2020	2059	01OCT2020	Low Utilization
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667	11AUG2020	2039	01NOV2020	High Utilization
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333	15SEP2020	2004	01DEC2020	High Utilization
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333	20OCT2020	1969	01JAN2021	Moderate Utilization
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333	12NOV2020	1946	01FEB2021	Low Utilization

Explanation

Machines are categorized based on utilization rate.

Utilization	Classification
≥ 90	High
70-89	Moderate
< 70	Low

Code 8 — Fraud Detection Macro

%macro fraud_check;

data fab_fraud_flag;

set fab_utilization;

length Fraud_Flag $20.;

if Yield_Percentage >100 then Fraud_Flag="Yield Error";

else if Downtime_Minutes <0 then Fraud_Flag="Downtime Error";

else if Utilization >100 then Fraud_Flag="Utilization Error";

else Fraud_Flag="Valid";

run;

proc print data=fab_fraud_flag;

run;

%mend;

%fraud_check;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours	Prod_Date	Days_Since_Production	Next_Service_Date	Util_Class	Fraud_Flag
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000	15FEB2020	2217	01MAY2020	Moderate Utilization	Valid
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000	10MAR2020	2193	01JUN2020	Moderate Utilization	Valid
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003	99.5	5.00000	05APR2020	2167	01JUL2020	High Utilization	Yield Error
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333	01MAY2020	2141	01AUG2020	Moderate Utilization	Downtime Error
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000	18JUN2020	2093	01SEP2020	Moderate Utilization	Valid
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333	22JUL2020	2059	01OCT2020	Low Utilization	Valid
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667	11AUG2020	2039	01NOV2020	High Utilization	Valid
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333	15SEP2020	2004	01DEC2020	High Utilization	Valid
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333	20OCT2020	1969	01JAN2021	Moderate Utilization	Valid
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333	12NOV2020	1946	01FEB2021	Low Utilization	Valid

Explanation

This macro detects suspicious values.

Possible fraud indicators:

·  Impossible yield values

·  Negative downtime

·  Utilization exceeding limits

Macros allow reusable validation logic.

Code 9 — Dataset Merge

data fab_cost;

input Fab_ID $ Production_Fees;

datalines;

FAB01 45000

FAB02 47000

FAB03 42000

FAB04 41000

;

run;

proc print data=fab_cost;

run;

OUTPUT:

Obs	Fab_ID	Production_Fees
1	FAB01	45000
2	FAB02	47000
3	FAB03	42000
4	FAB04	41000

proc sort data=fab_fraud_flag;by Fab_ID;run;

proc print data=fab_fraud_flag;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours	Prod_Date	Days_Since_Production	Next_Service_Date	Util_Class	Fraud_Flag
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000	15FEB2020	2217	01MAY2020	Moderate Utilization	Valid
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000	10MAR2020	2193	01JUN2020	Moderate Utilization	Valid
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003	99.5	5.00000	05APR2020	2167	01JUL2020	High Utilization	Yield Error
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333	01MAY2020	2141	01AUG2020	Moderate Utilization	Downtime Error
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000	18JUN2020	2093	01SEP2020	Moderate Utilization	Valid
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333	22JUL2020	2059	01OCT2020	Low Utilization	Valid
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667	11AUG2020	2039	01NOV2020	High Utilization	Valid
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333	15SEP2020	2004	01DEC2020	High Utilization	Valid
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333	20OCT2020	1969	01JAN2021	Moderate Utilization	Valid
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333	12NOV2020	1946	01FEB2021	Low Utilization	Valid

proc sort data=fab_cost;by Fab_ID;run;

proc print data=fab_cost;

run;

OUTPUT:

Obs	Fab_ID	Production_Fees
1	FAB01	45000
2	FAB02	47000
3	FAB03	42000
4	FAB04	41000

data fab_merged;

merge fab_fraud_flag 

      fab_cost;

by Fab_ID;

run;

proc print data=fab_merged;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours	Prod_Date	Days_Since_Production	Next_Service_Date	Util_Class	Fraud_Flag
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000	15FEB2020	2217	01MAY2020	Moderate Utilization	Valid
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000	10MAR2020	2193	01JUN2020	Moderate Utilization	Valid
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003	99.5	5.00000	05APR2020	2167	01JUL2020	High Utilization	Yield Error
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333	01MAY2020	2141	01AUG2020	Moderate Utilization	Downtime Error
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000	18JUN2020	2093	01SEP2020	Moderate Utilization	Valid
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333	22JUL2020	2059	01OCT2020	Low Utilization	Valid
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667	11AUG2020	2039	01NOV2020	High Utilization	Valid
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333	15SEP2020	2004	01DEC2020	High Utilization	Valid
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333	20OCT2020	1969	01JAN2021	Moderate Utilization	Valid
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333	12NOV2020	1946	01FEB2021	Low Utilization	Valid

Purpose

MERGE combines datasets horizontally.

Used when datasets share a common key.

Code 10 — Dataset Append

proc append base=fab_merged

            data=fab_cost force;

run;

proc print data=fab_merged;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours	Prod_Date	Days_Since_Production	Next_Service_Date	Util_Class	Fraud_Flag
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000	15FEB2020	2217	01MAY2020	Moderate Utilization	Valid
2	FAB02	WB1002	.	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000	10MAR2020	2193	01JUN2020	Moderate Utilization	Valid
3	FAB03	WB1003	0.55	105	10nm	300	110	42000	04-05-20	FAB03_WB1003	99.5	5.00000	05APR2020	2167	01JUL2020	High Utilization	Yield Error
4	FAB04	WB1004	0.40	89	7nm	-50	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333	01MAY2020	2141	01AUG2020	Moderate Utilization	Downtime Error
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000	18JUN2020	2093	01SEP2020	Moderate Utilization	Valid
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333	22JUL2020	2059	01OCT2020	Low Utilization	Valid
7	FAB07	WB1007	.	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667	11AUG2020	2039	01NOV2020	High Utilization	Valid
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333	15SEP2020	2004	01DEC2020	High Utilization	Valid
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333	20OCT2020	1969	01JAN2021	Moderate Utilization	Valid
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333	12NOV2020	1946	01FEB2021	Low Utilization	Valid
11	FAB01		.	.		.	.	45000			.	.	.	.	.
12	FAB02		.	.		.	.	47000			.	.	.	.	.
13	FAB03		.	.		.	.	42000			.	.	.	.	.
14	FAB04		.	.		.	.	41000			.	.	.	.	.

Purpose

APPEND adds observations to an existing dataset.

Code 11 — Dataset Transpose

proc transpose data=fab_fraud_flag out=fab_transposed;

var Yield_Percentage Utilization;

run;

proc print data=fab_transposed;

run;

OUTPUT:

Obs	_NAME_	COL1	COL2	COL3	COL4	COL5	COL6	COL7	COL8	COL9	COL10
1	Yield_Percentage	92	95	105	89	94	60	97	96	91	88
2	Utilization	85	88	110	80	75	40	92	95	70	65

Purpose

TRANSPOSE converts rows to columns.

Useful for report generation.

Code 12 — PROC DATASETS DELETE

proc datasets library=work nolist;

delete fab_clean_char fab_numeric fab_dates;

quit;

LOG:

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

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

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

Purpose

Removes intermediate datasets to keep workspace clean.

Final Corrected Dataset Code

data semiconductor_final;

set fab_utilization;

if Yield_Percentage >100 then Yield_Percentage=100;

if Downtime_Minutes <0 then Downtime_Minutes=0;

if Utilization >100 then Utilization=100;

if missing(Defect_Density) then Defect_Density=0.3;

run;

proc print data=semiconductor_final;

run;

OUTPUT:

Obs	Fab_ID	Wafer_Batch	Defect_Density	Yield_Percentage	Process_Node	Downtime_Minutes	Utilization	Production_Fees	Production_Date	Batch_Label	Effective_Yield	Downtime_Hours	Prod_Date	Days_Since_Production	Next_Service_Date	Util_Class
1	FAB01	WB1001	0.32	92	7nm	120	85	45000	02-15-20	FAB01_WB1001	88.8	2.00000	15FEB2020	2217	01MAY2020	Moderate Utilization
2	FAB02	WB1002	0.30	95	5nm	90	88	47000	03-10-20	FAB02_WB1002	.	1.50000	10MAR2020	2193	01JUN2020	Moderate Utilization
3	FAB03	WB1003	0.55	100	10nm	300	100	42000	04-05-20	FAB03_WB1003	99.5	5.00000	05APR2020	2167	01JUL2020	High Utilization
4	FAB04	WB1004	0.40	89	7nm	0	80	41000	05-01-20	FAB04_WB1004	85.0	-0.83333	01MAY2020	2141	01AUG2020	Moderate Utilization
5	FAB05	WB1005	0.28	94	5nm	60	75	39000	06-18-20	FAB05_WB1005	91.2	1.00000	18JUN2020	2093	01SEP2020	Moderate Utilization
6	FAB06	WB1006	0.90	60	14nm	500	40	35000	07-22-20	FAB06_WB1006	51.0	8.33333	22JUL2020	2059	01OCT2020	Low Utilization
7	FAB07	WB1007	0.30	97	5nm	40	92	48000	08-11-20	FAB07_WB1007	.	0.66667	11AUG2020	2039	01NOV2020	High Utilization
8	FAB08	WB1008	0.25	96	3nm	20	95	50000	09-15-20	FAB08_WB1008	93.5	0.33333	15SEP2020	2004	01DEC2020	High Utilization
9	FAB09	WB1009	0.45	91	7nm	200	70	43000	10-20-20	FAB09_WB1009	86.5	3.33333	20OCT2020	1969	01JAN2021	Moderate Utilization
10	FAB10	WB1010	0.60	88	10nm	350	65	42000	11-12-20	FAB10_WB1010	82.0	5.83333	12NOV2020	1946	01FEB2021	Low Utilization

This program fixes all data issues.

Corrections applied:

Issue	Correction
Missing defect density	replaced with average
Yield above 100	capped
Negative downtime	reset to 0
Utilization above 100	capped

Key Project Insights

1. Semiconductor fabs produce complex manufacturing datasets.
2. Data errors can significantly impact yield analysis.
3. SAS character functions standardize text variables.
4. Numeric functions help derive operational metrics.
5. Date functions allow production tracking.
6. Macros automate fraud detection.
7. Merge and append combine datasets.
8. Transpose helps in reporting.
9. PROC DATASETS manages workspace efficiently.
10. Data validation ensures reliable analytics.

Summary

This project demonstrates how Advanced SAS programming can be used to create, analyze, detect errors, and clean data related to semiconductor fabrication plants. The dataset represents operational information from semiconductor fabs, including variables such as Fab_ID, Wafer_Batch, Defect_Density, Yield_Percentage, Process_Node, Downtime_Minutes, Utilization, Production_Fees, and Production_Date. To simulate real-world manufacturing data challenges, several intentional errors were introduced into the raw dataset, such as missing defect density values, yield percentages greater than 100%, negative downtime values, and machine utilization exceeding 100%.

Using SAS programming techniques, the project demonstrates how to identify and correct these data issues systematically. Character functions like STRIP, TRIM, CATX, UPCASE, LOWCASE, and PROPERCASE were used to clean and standardize text variables. Numeric calculations were applied to derive useful metrics such as effective yield and downtime in hours. Date functions including MDY, INTCK, and INTNX were used to convert and analyze production dates and calculate future maintenance schedules.

The project also illustrates advanced dataset management techniques such as SET, MERGE, APPEND, and TRANSPOSE, along with workspace cleanup using PROC DATASETS DELETE. Additionally, a macro-based fraud detection logic was implemented to automatically flag suspicious or unrealistic values in the dataset.

Overall, the project highlights how SAS can be used to build reliable, error-free datasets for semiconductor manufacturing analytics, ensuring accurate operational insights and improved decision-making.

Conclusion

Semiconductor manufacturing is one of the most data-intensive industries. Reliable analytics requires accurate and validated datasets.

This project demonstrated how Advanced SAS programming techniques can be used to detect, analyze, and correct data errors in semiconductor fabrication plant datasets.

Using a combination of:

·  character functions

·  numeric calculations

·  date functions

·  macros

·  dataset manipulation techniques

SAS programmers can build robust data pipelines that ensure data quality and support high-value manufacturing decisions.

SAS INTERVIEW QUESTIONS

1. What is the difference between CALL SYMPUT and CALL SYMPUTX?

Answer:
CALL SYMPUT may include extra spaces, while CALL SYMPUTX removes leading and trailing spaces.

2. What is the purpose of %LET?

Answer:
%LET assigns values to macro variables.

3. What is %DO loop in macros?

Answer:
%DO loop repeats macro code multiple times.

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

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 SEMICONDUCTOR 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 

·  Clinical SAS Programmer

·  Research Data Analyst

·  Regulatory Data Validator

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

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 Data Processing and SQL-Driven Insights Reveal Operational Patterns in Indian Zoo Data Through Structured Reporting and Cross-Tabulation?

2.Can Advanced SAS Procedures Decode the Design, Popularity, and Impact of Futuristic Ganesh Mandaps Through Data-Driven Analysis?

3.How Can SAS Transform Raw Electronics Data into Smart Business Insights Using PROC SQL, Ranking Analytics, and Advanced Reporting Procedures?

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

About Us | Contact | Privacy Policy