Firdevs Ulus, Gurdal Ertek, Ozlem Kose, Gven Sahin

1
Visualizing the DEA

• Firdevs Ulus, Gurdal Ertek, Ozlem Kose, Güvenç
  Sahin

2
Our Study

• New framework
• Integrates some of the most popular approaches
  for benchmarking
• DEA, Financial Ratios, K-Means Clustering
• Helps understanding of how...
• to use these approaches and data visualization
  to get useful insights
• to compare and contrast these approaches with one
  another.

3
Outline

• Methodologies
• Dataset
• Framework
• Analyses Results
• Conclusion Future Work

4
Methodologies
5
Data Envelopment Analysis (DEA)

• Approach to measure efficiency of Decision
  Making Units (DMUs) in comparison to each other
• Uses data regarding the DMUs
• Inputs (Resources consumed)
• Outputs (Desired outcomes produced)
• Determination the efficient frontier,
  efficiency scores and reference sets

6
Financial Ratios

• Calculated based on
• Balance sheet
• Income statement.
• Used for
• comparing companies with respect to the
  standardized industry level
• evaluating a company in terms of its past and
  current performances.

7
Financial Ratios

• Turnover ratio
• ratio of the sales revenue to the net assets.
• Popular, because
• Is widely available
• Helps to make logical comparisons among companies
  in terms of their efficiencies

8
K-Means Clustering

• Cluster entities into k partitions based on their
  attributes
• Iteratively, nearest entities are grouped in
  the same clusters

9
Data Visualization

• Detecting outliers
• Finding patterns
• Coming up with new hypotheses
• Visualizations include
• Quantile plots, histograms, box plots, symmetry
  plots, scatter plots, quantile-quantile plots,
  etc.

10
Data Visualization
Colored scatter plot in Miner3D
www.miner3d.com
Tile Visualization in Visokio Omniscope
www.visokio.com/omniscope
Graph Visualization in yEd Graph Editor
www.yworks.com
11
Dataset

• Istanbul Chamber of Industry (ICI)
• ISO (ICI)-500 list of 2005
• 48 companies from food industry used and
  categorized into six groups
• milk products (M),
• flour (F)
• snack (S)
• poultry (P)
• oil/shortening (O)
• canned food (N)
• others (H).

12
Dataset

• Financial Data used in DEA K-means Clustering
• Equity
• Net Asset
• Sales from Production
• Sales Revenue
• Gross Added Revenue
• Profit from Operations
• Export
• Other useful data included in visual analyses
• Number of Workers

inputs
outputs
13
Framework
4 Types of analysis

• DEA Results Relevant Data from ISO 500
• DEA Results Financial Ratios
• DEA Results k-Mean Clustering Results
• Reference Sets According to the DEA Results

14
Analysis Results
15
4 Types of analysis

• DEA Results Relevant Data from ISO 500
• Analyses of DEA Results Financial Ratios
• Analyses of DEA Results k-Means Clustering
  Results
• Analysis of Reference Sets According to the DEA
  Results

16
DEA Results Relevant Data from ISO 500
17
4 Types of analysis

• DEA Results Relevant Data from ISO 500
• Analyses of DEA Results Financial Ratios
• Analyses of DEA Results k-Means Clustering
  Results
• Analysis of Reference Sets According to the DEA
  Results

18
DEA Results Relevant Data from ISO 500
19
4 Types of analysis

• DEA Results Relevant Data from ISO 500
• Analyses of DEA Results Financial Ratios
• Analyses of DEA Results k-Means Clustering
  Results
• Analysis of Reference Sets According to the DEA
  Results

20
DEA Results Financial Ratios
21
4 Types of analysis

• DEA Results Relevant Data from ISO 500
• Analyses of DEA Results Financial Ratios
• Analyses of DEA Results k-Means Clustering
  Results
• Analysis of Reference Sets According to the DEA
  Results

22
DEA Results k-Mean Clustering Results
23
4 Types of analysis

• DEA Results Relevant Data from ISO 500
• Analyses of DEA Results Financial Ratios
• Analyses of DEA Results k-Means Clustering
  Results
• Analysis of Reference Sets According to the DEA
  Results

24
Reference Sets According to the DEA Results
25
Conclusions

• Different benchmarking methodologies are not
  neccessarily consistent with each other.
• One should employ different methods of
  benchmarking, and interpret the results
• Our framework
• Encompasses the essentials of different
  benchmarking methods
• Enables an integrated comperative analysis

26
Future Work

• One can carry out formal statistical tests to
  prove or
• disprove the hypotheses
• that are claimed
• More financial ratio can be
• used to compare with the DEA results to get much
  more
• general results about these two methodologies
• performance.

27
(No Transcript)