Scribd
Upload
81 views

Code - Retail Analysis - Sas

The document analyzes sales data from multiple Walmart stores. It cleans and transforms the data into time series formats, calculates key metrics like growth rates and sales distributions. Models are developed to forecast demand for Store 1, including identifying best ARIMA parameters and producing a 4-period ahead forecast. Visualizations are created to provide monthly and semester sales insights by store.

Uploaded by

raviteja
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
81 views

Code - Retail Analysis - Sas

The document analyzes sales data from multiple Walmart stores. It cleans and transforms the data into time series formats, calculates key metrics like growth rates and sales distributions. Models are developed to forecast demand for Store 1, including identifying best ARIMA parameters and producing a 4-period ahead forecast. Visualizations are created to provide monthly and semester sales insights by store.

Uploaded by

raviteja
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

/* Import the Dataset */

FILENAME REFFILE '/folders/myfolders/Walmart_Store_sales.csv';

PROC IMPORT DATAFILE=REFFILE


DBMS=CSV
OUT=WORK.walmart replace;
GETNAMES=YES;
RUN;

proc print data=walmart;


run;

/* Check the content of the data */

proc contents data= work.walmart;


run;

/* Check the missing value */

proc means data=work.walmart nmiss;


run;

/* Which store has maximum sales */

proc means data=work.walmart max;


by store;
var weekly_sales;
run;

/* OR */

proc sort data=work.walmart;


by descending weekly_sales;
run;

/* Which store has maximum standard deviation */

proc summary data=work.walmart;


class store;
output out= walmart_standard(drop= _type_ _freq_) std(weekly_sales)=sd_max;
run;

proc print data=work.walmart_standard;


run;

proc sort data=work.walmart_standard;


by descending sd_max;
run;

proc print data=work.walmart_standard;


run;

/* Find out the coefficient of mean to standard deviation */

proc means data=work.walmart nonobs cv;


class store;
var weekly_sales;
run;

/* OR */

proc means data=work.walmart nonobs cv;


run;

/* Which store/s has good quarterly growth rate in Q3’2012 */

/* Filter year(2012) */

data date_12;
set work.walmart;
where year(date)=2012;
run;

proc print data= date_12;


run;

/* Calculate growth rate */

data growth;
format growth_rate percent8.2;
set work.date_12;
by store date weekly_sales;
lag_sales = ifn(first.store,0,lag(weekly_sales));
growth_rate = (weekly_sales/lag_sales)-1;
drop lag_sales;
run;

proc print data=growth;


run;

/* Convert the normat data into timeseries data */

proc timeseries data= growth out= good_growth;


by store;
id date interval= qtr accumulate=total;
var growth_rate;
run;

proc print data= good_growth;


run;

/* From timeseries data filterd only Q3 observations */

data good_growth_rate;
set good_growth;
where qtr(date)= 3;
run;

proc print data=good_growth_rate;


run;

/* Now Sort the data to see the good growth rate store wise */

proc sort data= good_growth_rate;


by descending growth_rate;
run;

proc print data= good_growth_rate;


run;

/* Some holidays have a negative impact on sales.


Find out holidays which have higher sales than
the mean sales in non-holiday season for all stores together */

/* Separate the holiday dates from main dataset's date */

data holiday;
set work.walmart;
where holiday_flag=1;
run;

proc print data=holiday;


run;

/* Separate the non-holiday dates from main dataset's date */

data non_holiday;
set work.walmart;
where holiday_flag=0;
run;
proc print data=non_holiday;
run;

/* Calculate the mean weekly_sales of the non-holiday data */

proc means data= non_holiday mean nonobs;


output out= mean_sales;
var weekly_sales;
run;

/* Compare the mean weekly_sales of the non-holiday data with weekly_sales of the holiday data */

proc sql;
create table holiday_sales as
select store, weekly_sales, date, holiday_flag as holiday,
case
when weekly_sales > 1041256.38 then 'Higher'
when weekly_sales < 1041256.38 then 'Lower'
end
as higher_sales
from holiday;
quit;

proc print data= holiday_sales;


run;

/* Finally found out holidays which have higher sales than


the mean sales in non-holiday season for all stores together */

data higher_holiday_sales;
set work.holiday_sales;
where higher_sales = 'Higher';
drop higher_sales;
title 'Higher Sales during Holidays';
run;

proc print data= higher_holiday_sales;


run;

/* Provide a monthly and semester view of sales in units and give insights */

/* Monthly view of sales in units */

/* Convert walmart data into timeseries data */

proc timeseries data= work.walmart


out= monthly_sales;
by store;
id date interval=month accumulate=total;
var weekly_sales holiday_flag temperature fuel_price cpi unemployment;
run;

proc print data=work.monthly_sales;


format weekly_sales dollar16.2;
run;

/* Giving insights */

/* Checking the correlation */

proc corr data= work.monthly_sales;


run;

/* 1. Doing Comparison */

/* a) Bar Chart */

proc sgplot data = work.monthly_sales;


hbar store/response = weekly_sales stat= sum
datalabel datalabelattrs=(weight=bold);
title 'Total Views by Store';
run;
/* b) Clustered Bar Chart / Column Chart */

data date;
set work.monthly_sales;
month = month(Date);
month_name=PUT(Date,monname.);
put month_name= @;
run;

proc print data= date;


run;

proc sgplot data= date;


vbar store/ response= weekly_sales group=month_name groupdisplay=cluster
datalabel datalabelattrs = (weight = bold) dataskin=gloss; yaxis grid;
title 'Total View by monthly wise';
run;

/* 2) Studying relationship */

/* a) Bubble Chart */

proc sgplot data = work.monthly_sales;


bubble X=weekly_sales Y=store size= weekly_sales
/fillattrs=(color = teal) datalabel = store;
run;

/* b) Scatter Plot for Relationship */

proc sgplot data = work.monthly_sales;


title 'Relationship of Store with Weekly_sales';
scatter X= weekly_sales Y = store/
markerattrs=(symbol=circlefilled size=15);
run;

/* 3. Studying Distribution */

/* a) Histogram */

proc sgplot data = work.monthly_sales;


histogram weekly_sales/fillattrs=(color = steel)scale = proportion;
density weekly_sales;
run;

/* b) Scatter Plot */

proc sgplot data = work.monthly_sales;


scatter X= date Y = weekly_sales/group= store groupdisplay=cluster
markerattrs=(symbol=circlefilled size=15);
run;

/* 4) Composition */

/* a) Stacked Column Chart: */

proc sgplot data= work.monthly_sales;


title 'Weekly_sales by Store and date';
vbar date / response= weekly_sales group= store stat=percent datalabel;
xaxis display=(nolabel);
yaxis grid label='Weekly_sales';
run;

/* Semester view of sales in units */

/* Convert walmart data into timeseries data */

proc timeseries data= work.walmart


out= semester_sales;
by store;
id date interval= semiyear accumulate= total;
var weekly_sales holiday_flag temperature fuel_price cpi unemployment;
run;
proc print data= work.semester_sales;
run;

/* Giving insights */

/* Checking the correlation */

proc corr data= work.semester_sales;


run;

/* 1. Doing Comparison */

/* a) Bar Chart */

proc sgplot data = work.semester_sales;


hbar store/response = weekly_sales stat= sum
datalabel datalabelattrs=(weight=bold);
title 'Total Views by Store';
run;

/* b) Clustered Bar Chart / Column Chart */

data date;
set work.semester_sales;
month = month(Date);
month_name=PUT(Date,monname.);
put month_name= @;
run;

proc print data= date1;


run;

proc sgplot data= date;


vbar store/ response= weekly_sales group=month_name groupdisplay=cluster
datalabel datalabelattrs = (weight = bold) dataskin=gloss; yaxis grid;
title 'Total View by monthly wise';
run;

/* 2) Studying relationship */

/* a) Bubble Chart */

proc sgplot data = work.semester_sales;


bubble X=weekly_sales Y=store size= weekly_sales
/fillattrs=(color = teal) datalabel = store;
run;

/* b) Scatter Plot for Relationship */

proc sgplot data = work.semester_sales;


title 'Relationship of Store with Weekly_sales';
scatter X= weekly_sales Y = store/
markerattrs=(symbol=circlefilled size=15);
run;

/* 3. Studying Distribution */

/* a) Histogram */

proc sgplot data = work.semester_sales;


histogram weekly_sales/fillattrs=(color = steel)scale = proportion;
density weekly_sales;
run;

/* b) Scatter Plot */

proc sgplot data = work.semester_sales;


scatter X= date Y = weekly_sales/group= store groupdisplay=cluster
markerattrs=(symbol=circlefilled size=15);
run;

/* 4) Composition */

/* a) Stacked Column Chart: */


proc sgplot data= work.semester_sales;
title 'Weekly_sales by Store and date';
vbar date / response= weekly_sales group= store stat=percent datalabel;
xaxis display=(nolabel);
yaxis grid label='Weekly_sales';
run;

/* For Store 1 – Build prediction models to forecast demand */

/* Store-1 data */

data store1;
set work.walmart;
where store = 1;
run;

proc print data= store1;


run;

/* Convert store-1 data into timeseries data */

proc timeseries data= store1


out= store_1;
by store;
id date interval= month accumulate= total;
var weekly_sales holiday_flag temperature fuel_price cpi unemployment;
run;

proc print data= work.store_1;


run;

/* Build Model */

ods noproctitle;
ods graphics / imagemap=on;

proc arima data=WORK.STORE_1 plots


(only)=(series(corr crosscorr) residual(corr normal)
forecast(forecast forecastonly) );
identify var=Weekly_Sales(1);
estimate p=(1 2 3) q=(1) method=ML;
forecast lead=4 back=0 alpha=0.05;
outlier;
run;

You might also like