• The given dataset is segregated on the basis of

day, month and time in order to understand

customer behaviour and trends.
• Earlier dataset consisted of 10 attributes. Where in
post churning of data, it consists of additional
attributes such as Day, Month and Time-Range.
• Day and Month are calculated based on Date
attribute.
• Time-Range for ease of understanding is
segregated as
• 4:00 am – 12:00 pm : Morning
• 12:01 pm – 16:00 pm : Afternoon
• 16:01 pm – 19:00 pm : Evening
• 19:01 pm – 00:00 pm : Night
• 00:01 pm – 3:59 am : Late Night/Early
Morning
• Dataset consists of 145830 records.
 70,000

• Through analysis of graphs and 63,987

charts one can identify night as 60,000

peak business hour which 50,000

gradually reduces though late
nights and early morning. 40,000

34,595
• Where in Morning experience 30,000
27,613
the lowest sales, which rises back
in afternoon and continues to 20,000
16,094
grow in evening, and again 10,000
peaks at night.
3,541
0
Night lateN/earlyM Morning Afternoon Evening
 Sales throughout the day
70,000

60,000

50,000

40,000

63,987
30,000

20,000
34,595
27,613
10,000
16,094

3,541
0
Night Morning lateN/earlyM Evening Afternoon
Customer Visit
 Business hours
1,60,00,000.00

1,40,00,000.00

1,20,00,000.00

1,00,00,000.00

80,00,000.00

60,00,000.00

40,00,000.00

20,00,000.00

0.00
Night Morning lateN/earlyM Evening Afternoon
Rate Tax Discount
 Sales VS Day and Time
30000

25000

20000

15000

10000

5000

0
Sat Fri Thru Wed Tue Mon Sun
Night Morning LateN/EarlyM Evening Afternoon
 Customer Orders

WINES WINES, 809

TOBACCO TOBACCO, 36496

MISC MISC, 1187

MERCHANDISE,
MERCHANDISE
487

LIQUOR &
LIQUOR & TOBACCO
TOBACCO, 54

LIQUOR LIQUOR, 6201

FOOD FOOD, 57023

BEVERAGE BEVERAGE, 43573

0 10000 20000 30000 40000 50000 60000

Customer Orders
WINES TOBACCO MISC
MERCHANDISE LIQUOR & TOBACCO LIQUOR
FOOD BEVERAGE

1%
15%

49%

27%

7%
0%
0%
1%
• Following items can be discontinued, since the sale is less than 100 quantity per year.
• Customer visit is high throughout the night i.e. 19:01 – 00:00 hours, which is followed by evening i.e. 16:01 – 19:00
hours.
• It’s been noticed that the least customer visits are through morning hours, i.e. 4:01 am to 12:00 pm.
• Weekend nights experience highest customer visits, i.e. Saturday and Sunday night.
• Customer visits drops during weekdays i.e. Monday – Thursday.
• Holiday season i.e. December and January have seen highest customer visit throughout the year.
• Summer season i.e. April, May and June has seen the least customer visit throughout the year.
• One can infer that December to January, Saturday and Sunday at nightly hours, the restaurant makes most of its
business.
• However one can also infer that during April to June, on weekdays specially during morning hours, customer visit is
the lowest.

• Recommendations
• One can recommend that Liquor & Tobacco category can be eliminated from the menu, since there’s already
Liqour and Tobacco individual categories, based on rate, tax and discount this could prove beneficial.
• One can also recommend discounts on Tobacco can be raised on weekdays for months between April to June.
• Discount of Liquor can be raised for months between April to June in order to increase customer head count.
• The below graph shows top 20 in demand items from different categories.
• It can be inferred that TOBACCO has most in demand and revenue generating items, in Beverages, Cappuccino is
the most demanded and ordered item.
• BEVERAGES • FOOD
• LIQUOR • LIQUOR & TOBACCO
• TOBACCO • WINES
• Merchandise • Miscellaneous
• Market Basket Analysis (MBA) is a data mining technique used to discover associations between items purchased
together in a transactional dataset. It is widely used in retail and e-commerce industries to understand customer purchasing
behavior and to optimize product placement, promotions, and cross-selling strategies.

• One of the key concepts in Market Basket Analysis is the notion of association rules. Association rules are statements
that describe the relationships between items in a dataset. They are typically written in the form of "if-then" statements,
where one set of items (the antecedent) implies the presence of another set of items (the consequent) in a transaction.

Relevance
• Identification and prediction with respect to high in demand combos can be identified using association rules.
• Item demands based on Time, Day and Month can be identified using association rules.
• Revenue generating Categories can be identified using association rules.
• Costing and discount over different section based on sales over time can be identified
• Support (Value used 0.0002)
• The support of an itemset is defined as the proportion of transactions in the dataset
in which the itemset appears. It indicates how frequently an itemset occurs together
in the dataset.
• Confidence (Value used 0.35)
• The confidence of a rule A -> B measures the proportion of transactions containing A
that also contain B. It reflects the likelihood that item B is purchased when item A is
purchased.
• Lift
• Lift measures how much more often the antecedent and consequent of a rule occur
together than would be expected if they were statistically independent. It's calculated
as the ratio of the observed support to the expected support if A and B were
independent.