408.Can We Build an Accurate Product Demand Forecasting & Fraud Detection System in SAS While Identifying and Fixing Intentional Errors?

Product Demand Forecasting and Fraud Detection Analytics Using SAS with Intentional Error Handling

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

HERE IN THIS PROJECT WE USED THESE SAS STATEMENTS —
DATA STEP | INPUT | DATALINES | FORMAT | SET | MERGE | IF-THEN-ELSE | LENGTH | STRIP | TRIM | CAT | CATX | PROPCASE | UPCASE | LOWCASE | COALESCE | ABS | MONTH | YEAR | MDY | INTNX | INTCK | PROC DATASETS DELETE | PROC SQL | PROC FREQ | PROC MEANS | PROC UNIVARIATE | PROC CORR | PROC TRANSPOSE  | PROC SGPLOT | SCATTER | REG | MACRO | %MEND | %UTILIZATION | %FRAUD_CHECK

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

Table of Contents

1.     Introduction

2.     Business Context

3.     Project Objectives

4.     Dataset Design Structure

5.     Raw Data Creation with Intentional Errors

6.     Error Identification & Explanation

7.     Corrected Full-Length SAS Code

8.     Data Transformation & Feature Engineering

9.     Forecast Accuracy Calculation

10.  Utilization Classification Macro

11.  Fraud Detection Logic Macro

12.  Statistical Analysis (PROC MEANS, UNIVARIATE, CORR)

13.  SQL-Based Analysis

14.  Visualization (PROC SGPLOT)

15.  Dataset Management (TRANSPOSE, DELETE)

16.  Why Each Procedure Is Used

17. 12 Key Points About The Project

18.  Conclusion

1. Introduction

Product demand forecasting is a critical analytics problem in supply chain management. Organizations rely on forecast models to estimate future sales demand and align procurement, production, and inventory accordingly.

However:

·       Forecasts may be inaccurate

·       Inventory may be underutilized or overstocked

·       Fraudulent sales patterns may distort actual data

·       Data quality issues may cause incorrect reporting

In this project, we simulate a real-world product demand forecasting environment using Base SAS, PROC SQL, statistical procedures, macros, and data management techniques.

We will:

·       Create a dataset (15+ observations)

·       Introduce intentional errors

·       Identify and correct them

·       Apply forecasting accuracy calculations

·       Perform fraud detection logic

·       Use date functions (MDY, INTCK, INTNX)

·       Apply numeric & character functions

·       Use TRANSPOSE, SET

·       Perform statistical and correlation analysis

·       Generate visualizations

2. Business Context

Assume a retail company operates across 4 regions:

·       North

·       South

·       East

·       West

Each month, the company:

·       Forecasts demand for products

·       Records actual sales

·       Tracks inventory levels

·       Calculates forecast error

·       Measures accuracy percentage

The company wants to answer:

·       Which region has high forecast error?

·       Which products are underutilized?

·       Is there suspicious sales inflation?

·       How accurate are forecasts?

·       Is inventory aligned with demand?

3.Dataset Structure

Variable	Description
Product_ID	Unique Product Code
Region	Sales Region
Forecasted_Demand	Predicted Units
Actual_Sales	Sold Units
Forecast_Error	Difference
Inventory_Level	Current Inventory
Accuracy_Score	Forecast Accuracy %
Sales_Date	Transaction Date
Fraud_Flag	Fraud Indicator

4. RAW DATA CREATION (WITH INTENTIONAL ERRORS)

data demand_raw;

 input Product_ID $ Region $ Forecasted_Demand Actual_Sales Inventory_Level 

      Sales_Date : date9.;

 format Sales_Date date9.;

 datalines;

P101 North 500 520 600 01JAN2025

P102 South 450 430 500 15JAN2025

P103 East 600 580 650 30JAN2025

P104 West 700 . 750 10FEB2025

P105 North 800 900 820 25FEB2025

P106 South -200 300 400 05MAR2025

P107 East 650 640 660 20MAR2025

P108 West 720 710 700 05APR2025

P109 North 550 0 580 15APR2025

P110 South 480 490 510 30APR2025

P111 East 620 610 630 10MAY2025

P112 West 710 730 720 20MAY2025

P113 North 900 950 910 05JUN2025

P114 South 500 10000 520 15JUN2025

P115 East 600 580 610 25JUN2025

;

run;

proc print data=demand_raw;

run;

OUTPUT:

Obs	Product_ID	Region	Forecasted_Demand	Actual_Sales	Inventory_Level	Sales_Date
1	P101	North	500	520	600	01JAN2025
2	P102	South	450	430	500	15JAN2025
3	P103	East	600	580	650	30JAN2025
4	P104	West	700	.	750	10FEB2025
5	P105	North	800	900	820	25FEB2025
6	P106	South	-200	300	400	05MAR2025
7	P107	East	650	640	660	20MAR2025
8	P108	West	720	710	700	05APR2025
9	P109	North	550	0	580	15APR2025
10	P110	South	480	490	510	30APR2025
11	P111	East	620	610	630	10MAY2025
12	P112	West	710	730	720	20MAY2025
13	P113	North	900	950	910	05JUN2025
14	P114	South	500	10000	520	15JUN2025
15	P115	East	600	580	610	25JUN2025

❗ Intentional Errors

·  Missing Actual_Sales (P104)

·  Negative Forecast (P106)

·  Zero Sales (P109)

·  Extremely High Sales (P114)

·  Forecast_Error not calculated

·  Accuracy not calculated 

What It Does

Creates the initial forecasting dataset with 15+ observations.

 Why Used

DATA step is used to manually create structured raw data.

 Business Meaning

Simulates real forecasting records including:

·       Forecast

·       Actual sales

·       Inventory

·       Sales date

Intentional errors were included:

·       Missing sales

·       Negative forecast

·       Extremely high sales

·       Zero sales

5. ERROR IDENTIFICATION

Missing Sales

proc means data=demand_raw n nmiss;

run;

OUTPUT:

The MEANS Procedure

Variable	N	N Miss
Forecasted_Demand Actual_Sales Inventory_Level Sales_Date	15 14 15 15	0 1 0 0

Outlier Detection

proc univariate data=demand_raw;

 var Actual_Sales;

run;

OUTPUT:

The UNIVARIATE Procedure

Variable: Actual_Sales

Moments
N	14	Sum Weights	14
Mean	1245.71429	Sum Observations	17440
Std Deviation	2530.79712	Variance	6404934.07
Skewness	3.68364406	Kurtosis	13.69523
Uncorrected SS	104989400	Corrected SS	83264142.9
Coeff Variation	203.160319	Std Error Mean	676.383981

Basic Statistical Measures
Location		Variability
Mean	1245.714	Std Deviation	2531
Median	595.000	Variance	6404934
Mode	580.000	Range	10000
		Interquartile Range	240.00000

Tests for Location: Mu0=0
Test	Statistic		p Value
Student's t	t	1.841726	Pr > |t|	0.0884
Sign	M	6.5	Pr >= |M|	0.0002
Signed Rank	S	45.5	Pr >= |S|	0.0002

Quantiles (Definition 5)
Level	Quantile
100% Max	10000
99%	10000
95%	10000
90%	950
75% Q3	730
50% Median	595
25% Q1	490
10%	300
5%	0
1%	0
0% Min	0

Extreme Observations
Lowest		Highest
Value	Obs	Value	Obs
0	9	710	8
300	6	730	12
430	2	900	5
490	10	950	13
520	1	10000	14

Missing Values
Missing Value	Count	Percent Of
		All Obs	Missing Obs
.	1	6.67	100.00

6. CORRECTED FULL-LENGTH DATA PROCESSING CODE

data demand_clean;

 set demand_raw;

 Product_ID = strip(upcase(Product_ID));

 Region = propcase(Region);

 if Forecasted_Demand < 0 then Forecasted_Demand = .;

 if Actual_Sales = . then Actual_Sales = 0;

 Forecast_Error = Actual_Sales - Forecasted_Demand;

 Accuracy_Score = (1 - abs(Forecast_Error)/Forecasted_Demand)*100;

 Month = month(Sales_Date);

 Year = year(Sales_Date);

 Month_Start = intnx('month', Sales_Date, 0, 'b');

 Month_End   = intnx('month', Sales_Date, 0, 'e');

 Months_Since_Start = intck('month', mdy(1,1,2025), Sales_Date);

 format Month_Start Month_End date9.;

run;

proc print data=demand_clean;

run;

OUTPUT:

Obs	Product_ID	Region	Forecasted_Demand	Actual_Sales	Inventory_Level	Sales_Date	Forecast_Error	Accuracy_Score	Month	Year	Month_Start	Month_End	Months_Since_Start
1	P101	North	500	520	600	01JAN2025	20	96.00	1	2025	01JAN2025	31JAN2025	0
2	P102	South	450	430	500	15JAN2025	-20	95.56	1	2025	01JAN2025	31JAN2025	0
3	P103	East	600	580	650	30JAN2025	-20	96.67	1	2025	01JAN2025	31JAN2025	0
4	P104	West	700	0	750	10FEB2025	-700	0.00	2	2025	01FEB2025	28FEB2025	1
5	P105	North	800	900	820	25FEB2025	100	87.50	2	2025	01FEB2025	28FEB2025	1
6	P106	South	.	300	400	05MAR2025	.	.	3	2025	01MAR2025	31MAR2025	2
7	P107	East	650	640	660	20MAR2025	-10	98.46	3	2025	01MAR2025	31MAR2025	2
8	P108	West	720	710	700	05APR2025	-10	98.61	4	2025	01APR2025	30APR2025	3
9	P109	North	550	0	580	15APR2025	-550	0.00	4	2025	01APR2025	30APR2025	3
10	P110	South	480	490	510	30APR2025	10	97.92	4	2025	01APR2025	30APR2025	3
11	P111	East	620	610	630	10MAY2025	-10	98.39	5	2025	01MAY2025	31MAY2025	4
12	P112	West	710	730	720	20MAY2025	20	97.18	5	2025	01MAY2025	31MAY2025	4
13	P113	North	900	950	910	05JUN2025	50	94.44	6	2025	01JUN2025	30JUN2025	5
14	P114	South	500	10000	520	15JUN2025	9500	-1800.00	6	2025	01JUN2025	30JUN2025	5
15	P115	East	600	580	610	25JUN2025	-20	96.67	6	2025	01JUN2025	30JUN2025	5

Explanation of Corrections:

1.STRIP + UPCASE

Removes extra spaces and standardizes product codes.

2.PROPCASE

Standardizes region names.

3.Negative Forecast Fix

Negative demand is unrealistic → converted to missing.

4.Missing Sales

Replaced with zero for safe calculation.

5. Forecast Error

Actual - Forecast.

6. Accuracy Formula

Industry standard forecast accuracy formula.

·  ABS( ) → Absolute error

·  MONTH( ) → Extract month

·  YEAR( ) → Extract year

7. UTILIZATION CLASSIFICATION MACRO

%macro utilization;

data demand_final;

 set demand_clean;

 Utilization_Ratio = Actual_Sales / Inventory_Level;

 length Utilization_Class $20;

 if Utilization_Ratio > 0.9 then Utilization_Class = "High";

 else if Utilization_Ratio > 0.7 then Utilization_Class = "Medium";

 else Utilization_Class = "Low";

run;

proc print data=demand_final;

run;

%mend;

%utilization;

OUTPUT:

Obs	Product_ID	Region	Forecasted_Demand	Actual_Sales	Inventory_Level	Sales_Date	Forecast_Error	Accuracy_Score	Month	Year	Month_Start	Month_End	Months_Since_Start	Utilization_Ratio	Utilization_Class
1	P101	North	500	520	600	01JAN2025	20	96.00	1	2025	01JAN2025	31JAN2025	0	0.8667	Medium
2	P102	South	450	430	500	15JAN2025	-20	95.56	1	2025	01JAN2025	31JAN2025	0	0.8600	Medium
3	P103	East	600	580	650	30JAN2025	-20	96.67	1	2025	01JAN2025	31JAN2025	0	0.8923	Medium
4	P104	West	700	0	750	10FEB2025	-700	0.00	2	2025	01FEB2025	28FEB2025	1	0.0000	Low
5	P105	North	800	900	820	25FEB2025	100	87.50	2	2025	01FEB2025	28FEB2025	1	1.0976	High
6	P106	South	.	300	400	05MAR2025	.	.	3	2025	01MAR2025	31MAR2025	2	0.7500	Medium
7	P107	East	650	640	660	20MAR2025	-10	98.46	3	2025	01MAR2025	31MAR2025	2	0.9697	High
8	P108	West	720	710	700	05APR2025	-10	98.61	4	2025	01APR2025	30APR2025	3	1.0143	High
9	P109	North	550	0	580	15APR2025	-550	0.00	4	2025	01APR2025	30APR2025	3	0.0000	Low
10	P110	South	480	490	510	30APR2025	10	97.92	4	2025	01APR2025	30APR2025	3	0.9608	High
11	P111	East	620	610	630	10MAY2025	-10	98.39	5	2025	01MAY2025	31MAY2025	4	0.9683	High
12	P112	West	710	730	720	20MAY2025	20	97.18	5	2025	01MAY2025	31MAY2025	4	1.0139	High
13	P113	North	900	950	910	05JUN2025	50	94.44	6	2025	01JUN2025	30JUN2025	5	1.0440	High
14	P114	South	500	10000	520	15JUN2025	9500	-1800.00	6	2025	01JUN2025	30JUN2025	5	19.2308	High
15	P115	East	600	580	610	25JUN2025	-20	96.67	6	2025	01JUN2025	30JUN2025	5	0.9508	High

What It Does

Classifies inventory utilization.

Why Used

Automates classification logic.

Business Meaning

Identifies:

·       Overstock (Low utilization)

·       Optimal stock (Medium)

·       Risk of stockout (High)

8. FRAUD DETECTION MACRO

%macro fraud_check;

data fraud_data;

 set demand_final;

 length Fraud_Flag $10;

 if Actual_Sales > 3*Forecasted_Demand then Fraud_Flag="Yes";

 else Fraud_Flag="No";

run;

proc print data=fraud_data;

run;

%mend;

%fraud_check;

OUTPUT:

Obs	Product_ID	Region	Forecasted_Demand	Actual_Sales	Inventory_Level	Sales_Date	Forecast_Error	Accuracy_Score	Month	Year	Month_Start	Month_End	Months_Since_Start	Utilization_Ratio	Utilization_Class	Fraud_Flag
1	P101	North	500	520	600	01JAN2025	20	96.00	1	2025	01JAN2025	31JAN2025	0	0.8667	Medium	No
2	P102	South	450	430	500	15JAN2025	-20	95.56	1	2025	01JAN2025	31JAN2025	0	0.8600	Medium	No
3	P103	East	600	580	650	30JAN2025	-20	96.67	1	2025	01JAN2025	31JAN2025	0	0.8923	Medium	No
4	P104	West	700	0	750	10FEB2025	-700	0.00	2	2025	01FEB2025	28FEB2025	1	0.0000	Low	No
5	P105	North	800	900	820	25FEB2025	100	87.50	2	2025	01FEB2025	28FEB2025	1	1.0976	High	No
6	P106	South	.	300	400	05MAR2025	.	.	3	2025	01MAR2025	31MAR2025	2	0.7500	Medium	Yes
7	P107	East	650	640	660	20MAR2025	-10	98.46	3	2025	01MAR2025	31MAR2025	2	0.9697	High	No
8	P108	West	720	710	700	05APR2025	-10	98.61	4	2025	01APR2025	30APR2025	3	1.0143	High	No
9	P109	North	550	0	580	15APR2025	-550	0.00	4	2025	01APR2025	30APR2025	3	0.0000	Low	No
10	P110	South	480	490	510	30APR2025	10	97.92	4	2025	01APR2025	30APR2025	3	0.9608	High	No
11	P111	East	620	610	630	10MAY2025	-10	98.39	5	2025	01MAY2025	31MAY2025	4	0.9683	High	No
12	P112	West	710	730	720	20MAY2025	20	97.18	5	2025	01MAY2025	31MAY2025	4	1.0139	High	No
13	P113	North	900	950	910	05JUN2025	50	94.44	6	2025	01JUN2025	30JUN2025	5	1.0440	High	No
14	P114	South	500	10000	520	15JUN2025	9500	-1800.00	6	2025	01JUN2025	30JUN2025	5	19.2308	High	Yes
15	P115	East	600	580	610	25JUN2025	-20	96.67	6	2025	01JUN2025	30JUN2025	5	0.9508	High	No

What It Does

Flags suspicious sales spikes.

Why Used

Detects abnormal patterns.

Business Meaning

Sales > 3x forecast may indicate:

·       Data manipulation

·       Reporting error

·       Fraud activity

9. SQL ANALYSIS

proc sql;

 create table summary_data as

 select Region,

       avg(Actual_Sales) as Avg_Sales,

       avg(Forecasted_Demand) as Avg_Forecast,

       avg(Accuracy_Score) as Avg_Accuracy

 from fraud_data

 group by Region;

quit;

proc print data=summary_data;

run;

OUTPUT:

Obs	Region	Avg_Sales	Avg_Forecast	Avg_Accuracy
1	East	602.5	617.500	97.545
2	North	592.5	687.500	69.486
3	South	2805.0	476.667	-535.509
4	West	480.0	710.000	65.265

10. PROC FREQ

proc freq data=fraud_data;

 tables Region*Fraud_Flag;

run;

OUTPUT:

The FREQ Procedure

Frequency Percent Row Pct Col Pct

Table of Region by Fraud_Flag Region Fraud_Flag No Yes Total East 4 26.67 100.00 30.77 0 0.00 0.00 0.00 4 26.67     North 4 26.67 100.00 30.77 0 0.00 0.00 0.00 4 26.67     South 2 13.33 50.00 15.38 2 13.33 50.00 100.00 4 26.67     West 3 20.00 100.00 23.08 0 0.00 0.00 0.00 3 20.00     Total 13 86.67 2 13.33 15 100.00

11. PROC MEANS

proc means data=fraud_data mean std min max;

 var Forecasted_Demand Actual_Sales Accuracy_Score;

run;

OUTPUT:

The MEANS Procedure

Variable	Mean	Std Dev	Minimum	Maximum
Forecasted_Demand Actual_Sales Accuracy_Score	627.1428571 1162.67 -53.0433682	129.6402461 2459.86 504.0110229	450.0000000 0 -1800.00	900.0000000 10000.00 98.6111111

12. PROC UNIVARIATE

proc univariate data=fraud_data;

 var Forecast_Error;

 histogram Forecast_Error;

run;

OUTPUT:

The UNIVARIATE Procedure

Variable: Forecast_Error

Moments
N	14	Sum Weights	14
Mean	597.142857	Sum Observations	8360
Std Deviation	2573.01508	Variance	6620406.59
Skewness	3.68631118	Kurtosis	13.7163928
Uncorrected SS	91057400	Corrected SS	86065285.7
Coeff Variation	430.887693	Std Error Mean	687.667205

Basic Statistical Measures
Location		Variability
Mean	597.1429	Std Deviation	2573
Median	-10.0000	Variance	6620407
Mode	-20.0000	Range	10200
		Interquartile Range	40.00000

Note: The mode displayed is the smallest of 2 modes with a count of 3.

Tests for Location: Mu0=0
Test	Statistic		p Value
Student's t	t	0.86836	Pr > |t|	0.4009
Sign	M	-1	Pr >= |M|	0.7905
Signed Rank	S	-1	Pr >= |S|	0.9657

Quantiles (Definition 5)
Level	Quantile
100% Max	9500
99%	9500
95%	9500
90%	100
75% Q3	20
50% Median	-10
25% Q1	-20
10%	-550
5%	-700
1%	-700
0% Min	-700

Extreme Observations
Lowest		Highest
Value	Obs	Value	Obs
-700	4	20	1
-550	9	20	12
-20	15	50	13
-20	3	100	5
-20	2	9500	14

Missing Values
Missing Value	Count	Percent Of
		All Obs	Missing Obs
.	1	6.67	100.00

The UNIVARIATE Procedure

13. PROC CORR

proc corr data=fraud_data;

 var Forecasted_Demand Actual_Sales Inventory_Level Accuracy_Score;

run;

 OUTPUT:

The CORR Procedure

4 Variables:	Forecasted_Demand Actual_Sales Inventory_Level Accuracy_Score

Simple Statistics
Variable	N	Mean	Std Dev	Sum	Minimum	Maximum
Forecasted_Demand	14	627.14286	129.64025	8780	450.00000	900.00000
Actual_Sales	15	1163	2460	17440	0	10000
Inventory_Level	15	637.33333	131.60909	9560	400.00000	910.00000
Accuracy_Score	14	-53.04337	504.01102	-742.60716	-1800	98.61111

Pearson Correlation Coefficients Prob > |r| under H0: Rho=0 Number of Observations
	Forecasted_Demand	Actual_Sales	Inventory_Level	Accuracy_Score
Forecasted_Demand	1.00000   14	-0.22556 0.4381 14	0.97736 <.0001 14	0.28222 0.3283 14
Actual_Sales	-0.22556 0.4381 14	1.00000   15	-0.18877 0.5004 15	-0.98561 <.0001 14
Inventory_Level	0.97736 <.0001 14	-0.18877 0.5004 15	1.00000   15	0.32319 0.2597 14
Accuracy_Score	0.28222 0.3283 14	-0.98561 <.0001 14	0.32319 0.2597 14	1.00000   14

What It Does

Calculates correlation matrix.

Why Used

Identifies relationships between variables.

14. TRANSPOSE

proc transpose data=summary_data out=transposed_summary;

 by Region NotSorted;

 var Avg_Sales Avg_Forecast Avg_Accuracy;

run;

proc print data=transposed_summary;

run;

OUTPUT:

Obs	Region	_NAME_	COL1
1	East	Avg_Sales	602.50
2	East	Avg_Forecast	617.50
3	East	Avg_Accuracy	97.55
4	North	Avg_Sales	592.50
5	North	Avg_Forecast	687.50
6	North	Avg_Accuracy	69.49
7	South	Avg_Sales	2805.00
8	South	Avg_Forecast	476.67
9	South	Avg_Accuracy	-535.51
10	West	Avg_Sales	480.00
11	West	Avg_Forecast	710.00
12	West	Avg_Accuracy	65.26

What It Does

Converts rows to columns.

Why Used

For reporting format change.

15. VISUALIZATION

proc sgplot data=fraud_data;

 scatter x=Forecasted_Demand y=Actual_Sales;

 reg x=Forecasted_Demand y=Actual_Sales;

run;

OUTPUT:

 What It Does

Creates scatter plot with regression line.

Why Used

Visual validation of forecast model.

16. PROC DATASETS – Clean Workspace

proc datasets library=work nolist;

 delete demand_clean demand_final fraud_data summary_data;

quit;

LOG:

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

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

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

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

Why Each Procedure Is Used

Procedure	Purpose
PROC SQL	Aggregation & summary
PROC FREQ	Categorical analysis
PROC MEANS	Summary statistics
PROC UNIVARIATE	Distribution & outlier
PROC CORR	Correlation
PROC SGPLOT	Visualization
PROC TRANSPOSE	Restructure data
PROC APPEND	Add datasets
PROC DATASETS DELETE	Clean workspace

17. 12 Key Points About THe Project

1. The project simulates a real-world retail demand forecasting environment across multiple regions, making it practical and business-oriented rather than purely theoretical.
2. A structured dataset was created with key variables such as Product_ID, Region, Forecasted_Demand, Actual_Sales, Inventory_Level, Forecast_Error, Accuracy_Score, Sales_Date, and Fraud_Flag to reflect real supply chain reporting systems.
3. Multiple intentional data issues were introduced, including missing sales values, negative forecasts, zero sales, and extremely high sales, to demonstrate error identification and correction techniques.
4. Data cleaning was performed using character functions like STRIP( ), UPCASE( ), and PROPCASE( ) to standardize product and region values and ensure consistent reporting.
5. Numeric validation logic was applied to handle unrealistic values, such as converting negative forecasts to missing and replacing missing sales with zero to prevent calculation failures.
6. Forecast performance was evaluated using calculated metrics such as Forecast_Error (Actual − Forecast) and Accuracy_Score percentage, providing measurable forecasting efficiency.
7. Date intelligence functions like MDY( ), INTNX( ), and INTCK( ) were used to derive monthly boundaries and time differences, enabling time-based analysis and reporting.
8. Inventory utilization was automated through a macro that classified products into High, Medium, and Low utilization categories based on the sales-to-inventory ratio.
9. Fraud detection logic was implemented using a macro to flag abnormal sales spikes where Actual_Sales significantly exceeded Forecasted_Demand.
10. Analytical procedures such as PROC SQL, PROC MEANS, PROC UNIVARIATE, PROC FREQ, and PROC CORR were used to generate summaries, distributions, correlations, and categorical insights.
11. Visualization using PROC SGPLOT helped validate the relationship between forecasted demand and actual sales through scatter plots and regression lines.
12. Dataset management techniques including TRANSPOSE, and PROC DATASETS DELETE ensured clean workflow execution, structured reporting, and scalable production-level processing.

18. Conclusion:

This project demonstrates how a structured SAS-based analytics workflow can transform raw forecasting data into meaningful business insights. By creating a realistic multi-region product demand dataset and intentionally introducing errors, the project highlights the importance of data validation, cleansing, and correction before analysis. Forecast error and accuracy calculations provide measurable indicators of model performance, while date functions enable time-based tracking. Inventory utilization classification supports stock optimization, and fraud detection logic identifies abnormal sales spikes that may indicate reporting issues or manipulation. Statistical procedures such as PROC MEANS, PROC UNIVARIATE, PROC FREQ, PROC CORR, and PROC SQL deliver descriptive, distributional, and relational insights. Visualization through PROC SGPLOT strengthens analytical interpretation. Overall, the project integrates data management, statistical evaluation, automation via macros, and business logic into a single scalable framework. It reflects real-world supply chain analytics practices and demonstrates strong  SAS programming competency suitable for enterprise-level forecasting and fraud monitoring environments.

SAS INTERVIEW QUESTIONS

·  How do you handle missing values?

·  What is reconciliation?

·  What is log checking?

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

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 FRAUD DETECTION SYSTEM 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 Do We Build a Complete River Trade and Pollution Monitoring System in SAS Using MERGE, APPEND, TRANSPOSE, and Macros?

2.How Can SAS Analyze Agricultural Machinery Performance Using DATA Step, PROC SQL, and Smart Date Calculations?

3.How Can SAS Analyze Different Types of Biryani Across the World Using PROC SQL, PROC FREQ, and PROC REPORT?

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

About Us | Contact | Privacy Policy