428.How Do SAS and R Complement Each Other in Detecting, Cleaning, and Transforming Complex Sensor Fusion Vehicle Data?

Advanced SAS and R Integration for Intelligent Error Detection and Transformation in Sensor Fusion Vehicle Data

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

SAS STATEMENTS USED

DATA STEP | LENGTH | INPUT | DATALINES | SET | MERGE | PROC SORT | PROC PRINT | PROC CONTENTS | PROC FREQ | PROC TRANSPOSE | PROC APPEND | PROC DATASETS DELETE | MACRO / %MACRO / %MEND | NUMERIC FUNCTIONS | CHARACTER FUNCTIONS

R EQUIVALENT STATEMENTS USED

data.frame() / tibble() | col_types / structure() | scan() / read.table() | inline data creation (c(), data.frame()) | rbind() / bind_rows() | merge() / dplyr::join functions | order() / dplyr::arrange() | print() / View() | str() / glimpse() | table() / count() | t() / pivot_longer() / pivot_wider() | bind_rows() | rm() | functions() | numeric functions (mean(), sum(), max()) | character functions (trimws(), toupper(), tolower(), paste(), paste0())

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

Introduction

Sensor fusion is a critical component in modern autonomous and semi-autonomous vehicles. It integrates data from multiple sensors such as LiDAR, Radar, and Cameras to make real-time driving decisions. However, real-world sensor data is noisy, inconsistent, and error-prone.

In this project, we will:

• Create a Sensor Fusion dataset
• Introduce intentional errors
• Detect and fix errors using SAS programming
• Apply data engineering techniques
• Use macros for fraud detection
• Perform transformations using:
• SET, MERGE, APPEND
• PROC TRANSPOSE
• Date functions (MDY, INTCK, INTNX)
• Character & numeric functions
• Mirror each step in R

Table of Contents

1. Business Context
2. Dataset Design
3. Raw Data Creation (with intentional errors)
4. Error Identification
5. Data Cleaning
6. Derived Variables
7. Date Handling
8. Utilization Classification
9. Dataset Merging
10. Appending Data
11. Transposing Data
12. Fraud Detection Macro
13. Functions Usage
14. Dataset Cleanup
15. Final Dataset
16. SAS vs R Key Points
17. Summary
18. Conclusion

Business Context

Automotive companies collect millions of sensor records daily. Problems include:

• Incorrect accuracy values (>100%)
• Missing sensor readings
• Invalid dates
• Fraudulent manipulation of performance metrics

Goal:
Build a robust data pipeline that detects and fixes these issues.

1. Create Raw Dataset (With Intentional Errors)

SAS Code

data sensor_raw;

    length Vehicle_ID $10 Date $12;

    input Vehicle_ID $ Lidar_Accuracy Radar_Accuracy Camera_Confidence 

          Detection_Latency Obstacle_Error_Rate Fees Utilization Date $;

    datalines;

V001 98 95 0.89 120 2.1 500 85 01-15-2024

V002 105 88 0.92 140 1.5 450 90 02-30-2024

V003 97 . 0.85 110 3.2 600 80 03-10-2024

V004 96 92 1.2 130 2.5 700 88 04-12-2024

V005 94 90 0.88 -10 2.0 550 75 05-20-2024

;

run;

LOG:

NOTE: The data set WORK.SENSOR_RAW has 5 observations and 9 variables.

R Code

sensor_raw <- data.frame(

  Vehicle_ID = c("V001","V002","V003","V004","V005"),

  Lidar_Accuracy = c(98,105,97,96,94),

  Radar_Accuracy = c(95,88,NA,92,90),

  Camera_Confidence = c(0.89,0.92,0.85,1.2,0.88),

  Detention_Latency = c(120,140,110,130,-10),

  Obstacle_Error_Rate = c(2.1,1.5,3.2,2.5,2.0),

  Fees = c(500,450,600,700,550),

  Utilization = c(85,90,80,88,75),

  Date = c("01-15-2024","02-30-2024","03-10-2024","04-12-2024","05-20-2024")

)

print(sensor_raw)

Errors Introduced

·  Lidar_Accuracy > 100 (invalid)

·  Radar_Accuracy missing

·  Camera_Confidence > 1

·  Negative latency

·  Invalid date (Feb 30)

2. Detect Errors

SAS Code

proc print data=sensor_raw;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85
2	V002	02-30-2024	105	88	0.92	140	1.5	450	90
3	V003	03-10-2024	97	.	0.85	110	3.2	600	80
4	V004	04-12-2024	96	92	1.20	130	2.5	700	88
5	V005	05-20-2024	94	90	0.88	-10	2.0	550	75

R Code

print(sensor_raw)

Why Used

·  Initial inspection

·  Helps identify anomalies

3. Fix Numeric Errors

SAS Code

data sensor_clean1;

    set sensor_raw;

    if Lidar_Accuracy > 100 then Lidar_Accuracy = 100;

    if Camera_Confidence > 1 then Camera_Confidence = 1;

    if Detection_Latency < 0 then Detection_Latency = .;

    if Radar_Accuracy = . then Radar_Accuracy = 85;

run;

proc print data=sensor_clean1;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75

R Code

sensor_clean1 <- sensor_raw

sensor_clean1$Lidar_Accuracy[sensor_clean1$Lidar_Accuracy > 100] <- 100

sensor_clean1$Camera_Confidence[sensor_clean1$Camera_Confidence > 1] <- 1

sensor_clean1$Detention_Latency[sensor_clean1$Detention_Latency < 0] <- NA

sensor_clean1$Radar_Accuracy[is.na(sensor_clean1$Radar_Accuracy)] <- 85

print(sensor_clean1)

Key Points

·  Data correction logic

·  Handling missing values

·  Ensures realistic ranges

4. Date Conversion (MDY)

SAS Code

data sensor_clean2;

    set sensor_clean1;

    Date_Converted = input(Date, mmddyy10.);

    format Date_Converted date9.;

run;

proc print data=sensor_clean2;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024

R Code

sensor_clean2 <- sensor_clean1

sensor_clean2$Date_Converted <- as.Date(sensor_clean2$Date, format="%m-%d-%Y")

print(sensor_clean2)

Why Important

·  Converts character to date

·  Required for time analysis

5. INTCK & INTNX

SAS Code

data sensor_dates;

    set sensor_clean2;

    Today = today();

    Days_Diff = intck('day', Date_Converted, Today);

    Next_Month = intnx('month', Date_Converted, 1);

    format Next_Month date9.;

run;

proc print data=sensor_dates;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024

R Code

library(lubridate)

sensor_dates <- sensor_clean2

sensor_dates$Today <- Sys.Date()

sensor_dates$Dates_Diff <- as.numeric(sensor_dates$Today - sensor_dates$Date_Converted)

sensor_dates$Next_Month <- sensor_dates$Date_Converted %m+% months(1)

print(sensor_dates)

Explanation

·  %m+% months(1) safely adds one month

·  Handles:

·  month-end dates

·  leap years

·  invalid rollover

·  INTCK → difference

·  INTNX → shifting dates

6. Character Functions

SAS Code

data sensor_char;

    set sensor_dates;

    Vehicle_ID_Clean = strip(Vehicle_ID);

    Vehicle_Upper = upcase(Vehicle_ID);

    Vehicle_Proper = propcase(Vehicle_ID);

    Combined = catx('-', Vehicle_ID, put(Lidar_Accuracy, 3.));

run;

proc print data=sensor_char;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month	Vehicle_ID_Clean	Vehicle_Upper	Vehicle_Proper	Combined
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94

R Code

sensor_char <- sensor_dates

sensor_char$Vehicle_ID_Clean <- trimws(sensor_char$Vehicle_ID)

sensor_char$Vehicle_Upper <- toupper(sensor_char$Vehicle_ID)

sensor_char$Vehicle_Proper <- tools::toTitleCase(sensor_char$Vehicle_ID)

sensor_char$Combined <- paste(sensor_char$Vehicle_ID, sensor_char$Lidar_Accuracy, sep="-")

print(sensor_char)

7. Utilization Classification

SAS Code

data sensor_class;

    set sensor_char;

    length Util_Class $8.;

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

    else if Utilization >= 80 then Util_Class="Medium";

    else Util_Class="Low";

run;

proc print data=sensor_class;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month	Vehicle_ID_Clean	Vehicle_Upper	Vehicle_Proper	Combined	Util_Class
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98	Medium
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100	High
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97	Medium
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96	Medium
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94	Low

R Code

sensor_class <- sensor_char

sensor_class$Util_Class <- ifelse(sensor_class$Utilization >= 90, "High",

                                   ifelse(sensor_class$Utilization >= 80, "Medium", "Low"))

print(sensor_class)

8. MERGE Example

SAS Code

data extra;

    input Vehicle_ID $ Fees_New;

    datalines;

V001 550

V002 480

;

run;

proc print data=extra;

run;

OUTPUT:

Obs	Vehicle_ID	Fees_New
1	V001	550
2	V002	480

proc sort data=sensor_class;by Vehicle_ID;run;

proc print data=sensor_class;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month	Vehicle_ID_Clean	Vehicle_Upper	Vehicle_Proper	Combined	Util_Class
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98	Medium
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100	High
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97	Medium
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96	Medium
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94	Low

proc sort data=extra;by Vehicle_ID;run;

proc print data=extra;

run;

OUTPUT:

Obs	Vehicle_ID	Fees_New
1	V001	550
2	V002	480

data merged;

    merge sensor_class 

          extra;

    by Vehicle_ID;

run;

proc print data=merged;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month	Vehicle_ID_Clean	Vehicle_Upper	Vehicle_Proper	Combined	Util_Class	Fees_New
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98	Medium	550
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100	High	480
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97	Medium	.
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96	Medium	.
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94	Low	.

R Code

extra <- data.frame(Vehicle_ID=c("V001","V002"), Fees_New=c(550,480))

print(extra)

merged <- merge(sensor_class, extra, by="Vehicle_ID", all=TRUE)

print(merged)

9. SET Statement

SAS Code

data combined;

    set sensor_class

        merged;

run;

proc print data=combined;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month	Vehicle_ID_Clean	Vehicle_Upper	Vehicle_Proper	Combined	Util_Class	Fees_New
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98	Medium	.
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100	High	.
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97	Medium	.
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96	Medium	.
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94	Low	.
6	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98	Medium	550
7	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100	High	480
8	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97	Medium	.
9	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96	Medium	.
10	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94	Low	.

R Code

Step 1: Add missing column

sensor_class$Fees_New <- NA

print(sensor_class)

Step 2: Match column order

sensor_class <- sensor_class[, names(merged)]

print(sensor_class)

Step 3: Apply rbind

combined <- rbind(sensor_class, merged)

print(combined)

Explanation

·  Makes both datasets structurally identical

·  Satisfies rbind( )  requirements

10. APPEND Statement

SAS Code

proc append base=sensor_class 

            data=extra force;

run;

proc print data=sensor_class;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month	Vehicle_ID_Clean	Vehicle_Upper	Vehicle_Proper	Combined	Util_Class
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98	Medium
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100	High
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97	Medium
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96	Medium
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94	Low
6	V001		.	.	.	.	.	.	.	.	.	.	.
7	V002		.	.	.	.	.	.	.	.	.	.	.

R Code

library(dplyr)

sensor_class <- bind_rows(sensor_class, extra)

Explanation

·  Matches columns by name

·  Adds missing columns

·  Fills with NA

11. TRANSPOSE

SAS Code

proc transpose data=sensor_class out=transposed;

    var Lidar_Accuracy Radar_Accuracy;

run;

proc print data=transposed;

run;

OUTPUT:

Obs	_NAME_	COL1	COL2	COL3	COL4	COL5	COL6	COL7
1	Lidar_Accuracy	98	100	97	96	94	.	.
2	Radar_Accuracy	95	88	85	92	90	.	.

R Code

transposed <- t(sensor_class[,c("Lidar_Accuracy","Radar_Accuracy")])

print(transposed)

12. Fraud Detection Macro

SAS Code

%macro fraud_check;

data fraud_flag;

    set sensor_class;

    if Lidar_Accuracy = 100 and Radar_Accuracy < 50 then Fraud="Yes";

    else Fraud="No";

run;

proc print data=fraud_flag;

run;

%mend;

%fraud_check;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month	Vehicle_ID_Clean	Vehicle_Upper	Vehicle_Proper	Combined	Util_Class	Fraud
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98	Medium	No
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100	High	No
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97	Medium	No
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96	Medium	No
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94	Low	No
6	V001		.	.	.	.	.	.	.	.	.	.	.						No
7	V002		.	.	.	.	.	.	.	.	.	.	.						No

R Code

sensor_class$Fraud <- ifelse(sensor_class$Lidar_Accuracy == 100 &

                                sensor_class$Radar_Accuracy < 50, "Yes","No")

print(sensor_class)

13. Numeric Functions

SAS Code

data numeric_calc;

    set sensor_class;

    Avg_Accuracy = mean(Lidar_Accuracy, Radar_Accuracy);

    Max_Acc = max(Lidar_Accuracy, Radar_Accuracy);

run;

proc print data=numeric_calc;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month	Vehicle_ID_Clean	Vehicle_Upper	Vehicle_Proper	Combined	Util_Class	Avg_Accuracy	Max_Acc
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98	Medium	96.5	98
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100	High	94.0	100
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97	Medium	91.0	97
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96	Medium	94.0	96
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94	Low	92.0	94
6	V001		.	.	.	.	.	.	.	.	.	.	.						.	.
7	V002		.	.	.	.	.	.	.	.	.	.	.						.	.

R Code

sensor_class$Avg_Accuracy <- rowMeans(sensor_class[,c("Lidar_Accuracy","Radar_Accuracy")], na.rm=TRUE)

sensor_class$Max_Acc <- apply(sensor_class[,c("Lidar_Accuracy","Radar_Accuracy")],1,max,na.rm=TRUE)

print(sensor_class)

14. PROC DATASETS DELETE

SAS Code

proc datasets library=work;

    delete sensor_raw extra;

quit;

LOG:

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

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

R Code

rm(sensor_raw, extra)

15. Final Corrected Dataset (Full Code)

SAS Code

data final_dataset;

    set sensor_class;

    if Lidar_Accuracy > 100 then Lidar_Accuracy = 100;

    if Camera_Confidence > 1 then Camera_Confidence = 1;

    Date_Converted = input(Date, mmddyy10.);

    format Date_Converted date9.;

    Avg_Accuracy = mean(Lidar_Accuracy, Radar_Accuracy);

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

    else if Utilization >= 80 then Util_Class="Medium";

    else Util_Class="Low";

run;

proc print data=final_dataset;

run;

OUTPUT:

Obs	Vehicle_ID	Date	Lidar_Accuracy	Radar_Accuracy	Camera_Confidence	Detection_Latency	Obstacle_Error_Rate	Fees	Utilization	Date_Converted	Today	Days_Diff	Next_Month	Vehicle_ID_Clean	Vehicle_Upper	Vehicle_Proper	Combined	Util_Class	Avg_Accuracy
1	V001	01-15-2024	98	95	0.89	120	2.1	500	85	15JAN2024	24186	796	01FEB2024	V001	V001	V001	V001-98	Medium	96.5
2	V002	02-30-2024	100	88	0.92	140	1.5	450	90	.	24186	.	.	V002	V002	V002	V002-100	High	94.0
3	V003	03-10-2024	97	85	0.85	110	3.2	600	80	10MAR2024	24186	741	01APR2024	V003	V003	V003	V003-97	Medium	91.0
4	V004	04-12-2024	96	92	1.00	130	2.5	700	88	12APR2024	24186	708	01MAY2024	V004	V004	V004	V004-96	Medium	94.0
5	V005	05-20-2024	94	90	0.88	.	2.0	550	75	20MAY2024	24186	670	01JUN2024	V005	V005	V005	V005-94	Low	92.0
6	V001		.	.	.	.	.	.	.	.	.	.	.					Low	.
7	V002		.	.	.	.	.	.	.	.	.	.	.					Low	.

R Code

final_dataset <- sensor_class

final_dataset$Lidar_Accuracy[final_dataset$Lidar_Accuracy > 100] <- 100

final_dataset$Camera_Confidence[final_dataset$Camera_Confidence > 1] <- 1

final_dataset$Date_Converted <- as.Date(final_dataset$Date, format = "%m-%d-%Y")

final_dataset$Avg_Accuracy <- rowMeans(final_dataset[,c("Lidar_Accuracy","Radar_Accuracy")], na.rm=TRUE)

final_dataset$Util_Class <- ifelse(final_dataset$Utilization >= 90,"High",

                                   ifelse(final_dataset$Utilization >= 80,"Medium","Low"))

print(final_dataset)

SAS vs R — 15 Key Points

1. Data Ingestion

·       SAS: DATA step, INFILE, PROC IMPORT provide structured ingestion with strong typing

·       R: read.csv( ), readr, data.table::fread( ) offer flexible but less strict ingestion

2. Schema Enforcement

·       SAS: Fixed metadata (length, type) ensures consistency

·       R: Dynamic typing; schema inconsistencies may silently propagate

3. Handling Missing Values

·       SAS: Missing numeric = . handled uniformly across procedures

·       R: Uses NA, requires explicit handling (na.rm=TRUE)

4. Error Detection (Outliers)

·       SAS: IF conditions, PROC UNIVARIATE, PROC MEANS

·       R: summary(), boxplot.stats(), dplyr::filter()

5. Data Validation

·       SAS: Strong rule-based validation in DATA step

·       R: Requires manual checks or packages like validate, assertthat

6. Data Correction

·       SAS: Inline correction using IF-THEN logic in DATA step

·       R: Vectorized replacement using indexing or mutate()

7. Date Handling

·       SAS: MDY, INTCK, INTNX are built-in and robust

·       R: Base date + lubridate needed for equivalent flexibility

8. Row-wise Processing

·       SAS: Native row-by-row execution (DATA step)

·       R: Vectorized; row-wise requires apply( ) or rowwise( )

9. Dataset Appending

·       SAS: SET automatically aligns variables

·       R: rbind( ) fails unless structure matches; bind_rows( ) needed

10. Merging Datasets

·       SAS: MERGE BY with sorted datasets

·       R: merge( ) or dplyr::left_join( ) more flexible

11. Transpose Operations

·       SAS: PROC TRANSPOSE (simple, structured)

·       R: t( ), pivot_longer( ), pivot_wider( ) (more flexible but verbose)

12. String Handling

·       SAS: Functions like STRIP, TRIM, UPCASE, PROPCASE

·       R: trimws( ), toupper( ), tolower( ), stringr package

13. Numeric Computation

·       SAS: MEAN, SUM, MAX handle missing automatically

·       R: Requires na.rm=TRUE explicitly

14. Automation (Macros vs Functions)

·       SAS: Macro language (%MACRO) for dynamic code generation

·       R: Functions and loops; more powerful but less declarative

15. Error Logging & Debugging

·       SAS: Built-in log with warnings, notes, errors

·       R: Console-based; debugging requires traceback( ), debug( )

Summary

Both SAS and R are capable of detecting, analyzing, and fixing errors in sensor fusion datasets, but their approaches differ fundamentally:

·       SAS is:

o   Structured

o   Metadata-driven

o   Ideal for regulated environments (like clinical trials, automotive safety logs)

o   Strong in data validation, traceability, and reproducibility

·       R is:

o   Flexible

o   Developer-driven

o   Powerful for custom analytics and machine learning

o   Requires careful handling of structure and missing values

For sensor fusion data (like LiDAR, Radar, Camera metrics):

·       SAS ensures data integrity and compliance

·       R enables advanced analytics and modeling

Conclusion

Advanced SAS programming provides a robust, rule-based framework for detecting and correcting errors in structured datasets such as sensor fusion vehicle data. Its strengths lie in consistency, automatic handling of structure, and built-in validation mechanisms.

In contrast, R offers greater flexibility and analytical power, but demands stricter developer discipline for handling inconsistencies, especially in real-world messy data.

Best Practice in Industry:

·       Use SAS for:

o   Data cleaning

o   Regulatory reporting

o   Structured pipelines

·       Use R for:

o   Advanced analytics

o   Visualization

o   AI/ML modeling

SAS INTERVIEW QUESTIONS

1. What is the difference between macro variables and dataset variables?

Answer:
Macro variables exist during program compilation, while dataset variables exist in datasets during execution.

2. 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.

3. What is the purpose of %LET?

Answer:
%LET assigns values to macro variables.

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

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 VEHICLE 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 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