NARSEE MONJEE INSTITUTE OF MANAGEMENT STUDIES

SCHOOL OF BUSINESS MANAGEMENT

Business Analytics

Group Assignment

FTMBA, Trimester 2, 2021-22

Submitted To Faculty: Prof. Mukund Prasad

Submitted By: Group 6 – Division D – I Year

Team Member SAP ID Roll No

Mohit Agarwal 80512100727 D006

Abhay Kumar Singh 80512100251 D016

Prakriti Tiwari 80512100588 D017

Pranjal Verma 80512100238 D046

Amber Goel 80512100735 D056

1
Global Retail Industry Introduction

Global retail market is driven by burgeoning population, rising GDP growth, greater disposable
income, and increasing consumer spending. The retail industry revolves around commodities for
personal or household consumption like food, apparel and accessories, home improvement,
technology, pharmaceuticals, beverages, etc. Asia Pacific (APAC) dominates the industry, comprising
of 35% of the global market.
The global economy slowed in 2019 and due to COVID-19 outbreak crashed in 2020 creating major
challenges for the retail industry. The retail industry is expected to show slow recovery since most of
the products in the industry are sensitive to economic cycles.

Figure 1: Growth in Retail

Use Cases of Retail Analytics

 Sales profitability and demand forecasting: Sales analytics gives insights into complete
picture of each transaction which can help in improving sales and overall profitability
 Supply chain analytics and inventory management: These systems help in developing
robust demand forecasts through a statistical analysis of data across outlets. This helps in
giving a seamless experience to the customers.
 New Product release plan: Identification of ideal locations for product launches, features
most desirable by the customers, or price ranges at which the product would sell best can all
be effortlessly analyzed through retail analytics.
 E-commerce analytics: The power of heatmaps, recommendation engines, fraud detection,
and customer lifetime value analytics provides insights to enhance the end-user experience.
 Promotions: Promotions are customer magnets and are cleverly targeted to generate
greater sales. Analytics allows the study of deeper consumer behavior. Predictive Analytics

2
 monitors the actions of consumers minutely to dish out a corresponding promotion by
aligning them with the buying needs of the consumer.
 Behavior Analytics: Understanding consumer behaviour to further work on communications
with them to increase revenue and decrease the acquisition cost are some of the main
challenges faced by retailers. Predictive Analytics is a data-driven technique that filters
information to gather useful information and understand the consumers better.
 Fraud Detection: Retailers revenue has been severely affected by fraud detection. Fraud
damages the reputation of firms and retailers when consumers face difficulty. Predictive
Analytics helps protect against fraudsters by proactive identification.

Implementation of Dashboard Creation

Step 1: Data uploaded to Microsoft SSMS
We selected Adventure works as a data source of choice. Adventure Works is a multinational retail
chain which deals in bikes and other sport equipment. We download a screenshot of the database
backup and uploaded it to Microsoft SSMS database. 

Figure 2: ER Diagram of Adventure Works Database

3
Step 2: Establish connection with power query 
The database is connected to power query which is a data transformation utility of PowerBI. We
have used this utility to add columns, perform row level calculations and change data types.

Figure 3: Transformation in Power Query

Step 3: Data Modelling

We used the data modelling tab to define relationships between tables by defining joins. We also
define the preferred aggregation method of various measures and they display styles like currency
on this tab.

Figure 4: Data Model

4
Step 4: Define measures and create visualizations

Figure 5: Visualization Creation on PowerBI

Analysis and Implication of analysis 

Screen 1: Leadership Sales Dashboard
Audience: Senior Leaders of the organization 
Need: A quick flash report which showcases quick numbers like sales, transactions, average sales.
Analysis Displayed 
1. Sales: KPI card visual displays the total amount of sales for the selected month/duration.
2. Transactions: Count of transactions for the selected month 
3. Average Sales: This metric gives the average sales per transaction. Helps to gauge how much
the buyers are spending on each transaction.
4. Sales and Sales LY: This bar chart visually displays the comparison of sales and Sales LY for
different territories. The leader can compare the performance of different locations wrt each
other and versus their last year revenue.
5. Sales over the years: This visualization shows the comparison of sales with last year's sales.
Helps the leaders see if the company is progressing as usual or their any outliers.
6. Sales by Product Category: This is a pie chart showing the sales value contribution of
different product categories.
7. Delay analysis: This is a bar chart that shows the count of transactions that have been
delayed. This is a critical KPI from a leadership standpoint as customer experience is directly
affected.
8. Top 10 Selling and Bottom 10 Selling Products: These grids help the CXO see what products
are doing well and what are not doing well.

5
 Figure 6: Screen 1 Leadership Sales Dashboard

Insights we can derive

1. Overall Performance: This screen helps form an overall picture of the performance of the
organisation. A reflection of organisation policy that might have been taken recently.
2. Comparative performance: Helps identify if any particular territory is lagging and specific
action needs to be taken.
3. Product Identification:  Identifying the products that are doing well and products that are
not doing well. This helps decision makers take decisions on new product launches,
discontinuing products.

6
Screen 2: Regional Analysis
Audience: Regional Management, Marketing Professionals
Need: A quick flash report which showcases quick numbers helping identify regional performance.

Figure 7: Screen 2 Regional Analysis

Analysis Displayed 
1. Region-wise Sales: A map visualization showing sales of various territories.
2. Product Category Sales: Compares the sales of various product categories across locations. 
3. Top Promotions by sales/Top Promotions by transaction count:  This enlists the top
promotions. They can be filtered by territory.
4. Top Performing Employee by territory: Helps identify the most successful employee in the
territory.
Insights we can derive
1. Overall Regional Performance: This screen helps form an overall picture of the performance
of the Region and territories. A reflection of regional performance.
2. Comparative performance: Helps identify if any particular territory is lagging and specific
action needs to be taken. Insights around which category is doing well in which territory.
3. Promotion Identification:  Identifying the Promotions that are doing well and the
contribution of sales that are arising due to the promotions. This helps decision-makers take
decisions on new promotions, discontinuing promotions.
4. Identify Top Performers: Help identify the top performers so that they can be rewarded and
appreciated for their contributions.

7
Screen 3: Market Basket Analysis
Audience: Marketing and Merchandising Department 
Need: The Basket Analysis uncovers the goods that customers bought together.
This research isn't just about products. The basket analysis may also be used to investigate the most
frequently viewed pages of a website, the most frequently asked questions in support requests, and
even the foods bought at a restaurant. We for this project demonstration have used product-
subcategory as the granular level. 

Analysis Displayed 
Three key concepts required to perform basket analysis: Support, Confidence, and Lift.

Figure 8: Screen 3 Market Basket Analysis

Support
It's the proportion of transactions that contain two distinct items. If two goods appear together in
0.9 percent of transactions, they have 0.9 percent support.
Support is vital for two reasons: 
– The level of support should be sufficient to make the cross-selling possibility financially viable.
Cross-selling items with low profit margins is usually not worth the time and effort.
– Unusual product combinations (poor support) should not be employed since statistical analysis on
rare combinations is unreliable.

8
Support = Number of Transactions including one or Multiple Products/Total number of
transactions
Dax Formula
Support basket 1 =
var item1= [Name]
var item2=[Item2]
var transactionsWithItem1 = SELECTCOLUMNS(FILTER('marketbasket',marketbasket[Name]=
item1),"transaction ID",marketbasket[CustomerID])
var transactionsWithItem2 = SELECTCOLUMNS(FILTER('marketbasket',marketbasket[Name]=
item2),"transactionID",marketbasket[CustomerID])
var transactionsWithBothItems = INTERSECT(transactionsWithItem1,transactionsWithItem2)
RETURN COUNTROWS(transactionsWithBothItems)/ DISTINCTCOUNT(marketbasket[CustomerID])

Confidence
Confidence is the percentage of transactions that contain the two products together, out of the
transactions containing one of the two products.
Confidence helps us determine the direction of the cross-selling. 
Confidence does not give us the strength of the relationship between the two products. Confidence
in isolation can highlight incorrect information. 
Confidence of product one-> Basket = Support of Basket/Support of Product one
Confidence of product two-> Basket = Support of Basket/Support of Product two
Dax Formula
Confidence Item 1 -> Basket =
var item1 = 'Basket analysis'[Name]
var numberOfTransactions = DISTINCTCOUNT(marketbasket[CustomerID])
var supportItem1 = COUNTROWS(FILTER('marketbasket',marketbasket[Name]=item1))/
numberOfTransactions
RETURN [Support basket 1]/supportItem1

Lift
The strength of the link between two goods in a basket is measured by lift.

 A lift less than 1 indicates that there is no important relationship between the products.
 A number greater than one (a positive connection) shows that customers purchase the
basket more frequently than would occur by chance.
 A number less than one (a negative connection) suggests that customers purchase the
basket less frequently than would occur by chance.

9
If two goods have a Lift greater than a threshold, people are more likely to buy them together than if
they were purchased separately. This is a really strong connection. We have given a slicer for end
users to select the value of lift.
Lift = Support of basket / ((Support of product one * Support of product two)
Lift =
var item1 = 'Basket analysis'[Name]
var item2 = [Item2]
var numberOfTransactions = DISTINCTCOUNT(marketbasket[CustomerID])
var supportItem1 =
COUNTROWS(FILTER(marketbasket,marketbasket[Name]=item1))/numberOfTransactions
var supportItem2 =
COUNTROWS(FILTER('marketbasket','marketbasket'[Name]=item2))/numberOfTransactions
RETURN [Support basket 1]/(supportItem1*supportItem2)

Insights we can derive

1. Help us form combos to encourage cross selling 
2. Merchandising solutions can remind people what they need to shop and also improve
consumer experience
3. Optimize self-spaces 

Conclusion
In this report, we analyzed the data of the Adventure Works store using Power BI as our data
analytics tool. This helped us in analyzing overall regional performance, comparative performance,
Identification of successful products and promotions, and recognition of top performers through the
data captured. We also performed market basket analysis using Support, Confidence, and Lift. In the
future, we can use this tool for fraud detection, consumer behavior analytics, and supply chain
analytics as well.

References
 https://www.mordorintelligence.com/industry-reports/retail-industry
 https://www.lucintel.com/retail-market-2017.aspx
 https://www.polestarllp.com/top-5-ecommerce-retail-analytics-use-cases
 https://farrago.ai/top-10-use-cases-for-predictive-analytics-in-retail/
 Avi Singh - PowerBIPro - YouTube
 Power BI: Basket Analysis Full Tutorial - Finance BI (finance-bi.com)  

10