Operation Data Analysis Hints and Guidelines

1
Operation Data Analysis Hints and Guidelines
EIN 6133 Enterprise Engineering Summer, 2015
2
Tools to Analyze Data

• Tools to analyze data range from simple to
  complex
• Reports and graphs
• Advanced statistics forecasting models
• Advanced optimization models and tools
• Having the right people matters
• Having data modeling

3
A Large Quantity of Quality Data

• All analytic methods feeds on data in large
  quantity and good quality
• Having good data can be turned into a competitive
  advantage
• Integrated organizations have a lot of data
  available, they must learn to exploit it

4
Interpreting Data

• Skills are required to
• create appropriate graphs, reports, and
  statistical analysis
• interpret correctly graphs, reports and
  statistics, and
• make appropriate decisions from the analytics

5
Using Queries to Analyze Data
A primary key is an attribute, or a combination
of attributes, that identify in a unique way each
row in a table. A primary key has to always
contain a value. It cannot be empty.
A foreign key is an attribute of a table (that
can be composed), meanwhile, the foreign key is a
primary key of another table. These two tables
are linked each to other by the foreign key.
There is an important concept associated with a
foreign key referential integrity. We say that a
relation has referential integrity if all the
values of a foreign key attribute in a table
exist in the table where this attribute is a
primary key. An example will be shown later.
6
Using Queries to Analyze Data
7
Using Queries to Analyze Data
A logical data model of a 1 to N relationship
8
Using Queries to Analyze Data
9
Using Queries to Analyze Data

• Queries contain 2 basic elements
• Key Figures, KPI
• Dimensions. (Characteristics)

Margins as a function of time
Sales by country
10
An Example
Dimensions
Dimensions
Measures
11
Elements of an Info Cube

• Key figures
• Dimensions

12
Types of Measures

• Additive  it makes sense to sum the measures
  across all dimensions
• Quantity sold across Region, Store, Salesperson,
  Date, Product
• semi additive  additive only across certain
  dimensions
• Quantity on hand is not additive over Date, but
  it is additive across Store and Product
• non additive  cannot be summed across any
  dimensions
• A ratio, a percentage
• A measure that is non additive on one dimension
  may be the object of other data aggregations
• Average, Min, Max of quantities on hand over time

13
How DW Differs from a Transactional DB?
Characteristic DB DW
Operation Real-time, transactional Decision support, strategic analysis
Model Entity-Relationship Star Schema
Redundant data Designed to avoid Permitted
Data Raw data, current Aggregated, Historical data,
of users Many Few
Update Immediate Deferred
Calculated fields None stored Many stored
Mental model Tabular Hypercube
Queries Simple, some saved Complex, many saved
Operations Read / Write Read Only
Size Go (Gigabytes) To(Terabytes)
14
Exploring Data
15
Plant B an Overview
16
Plant C an Overview
17
Try to Maintain Stocks for All Products
18
Large Variations in Sales per Step
19
Manipulating Graphs
20
Graph type Scattered Bars
21
Graph Type Lines
22
Graph Type 3D Bars
22
23
Questions
24
Question 1

• Current assets include
• (i) cash
• (ii) receivables
• (iii) raw material inventory (for mfg game)
• (iv) finished product inventory
• How well have the teams performed in managing the
  current assets over time?
• Hint Use the financial data

24
25
Question 2

• Did the winning team bring their highest margin
  product to market first?
• Did they charge a price premium while they were
  first to market?
• Can you see the impact of a competitor entering
  the market?
• Hint Use the operational data

25
26
Question 3

• One objective of materials management is to make
  sure that raw materials are available for
  production when needed
• Which company has managed this process well as
  shown by having the largest variety of products
  in stock?
• Hint Use inventory data by products over time

26
27
Question 4

• Companies may have different strategies for
  production management
• Some may prefer long productions to minimize
  setup losses, while others may prefer shorter
  runs to respond more quickly to market
  opportunities
• Can you determine what strategies were used by
  each team?
• Where there any production disruptions?
• Hint Use production data over time and
  products. Filter for each individual company.

27
28
Question 5

• Companies want to maximize sales
• If sales are too high, the price may be too low,
  and vice versa
• Can you tell sales is affected by prices?

28
29
Question 6

• Who owns the market (as measured by market share)
  for each product?
• Hint Use sales data filtered by product with
  drilldown across plant
• Use a stacked area chart

29