Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data...

12
Supplementary Material S.1. Instructions to Access the Original Dataset Data Publication http://pubsonline.informs.org/doi/abs/10.1287/msom.1070.0176 Data Listing from Original Publisher http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176 Individual Data Items Dataset Description http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176/suppl_file/ msom.1070.0176-sm-data_set.pdf Dataset in Microsoft Access Format http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176/suppl_file/ msom.1070.0176-sm-datainaccess.zip Dataset in Microsoft Excel Format http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176/suppl_file/ msom.1070.0176-sm-datainexcel.zip Dataset in XML Format http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176/suppl_file/ msom.1070.0176-sm-datainxml.zip S-1

Transcript of Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data...

Page 1: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

Supplementary Material

S.1. Instructions to Access the Original Dataset

• Data Publication

http://pubsonline.informs.org/doi/abs/10.1287/msom.1070.0176

• Data Listing from Original Publisher

http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176

• Individual Data Items

– Dataset Description

http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176/suppl_file/

msom.1070.0176-sm-data_set.pdf

– Dataset in Microsoft Access Format

http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176/suppl_file/

msom.1070.0176-sm-datainaccess.zip

– Dataset in Microsoft Excel Format

http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176/suppl_file/

msom.1070.0176-sm-datainexcel.zip

– Dataset in XML Format

http://pubsonline.informs.org/doi/suppl/10.1287/msom.1070.0176/suppl_file/

msom.1070.0176-sm-datainxml.zip

S-1

Page 2: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

S.2. Pseudocode for Network Layouts

To enhance repeatability and relevance, we provide pseudo-codes for implementing five network

layouts used in this paper. Our pseudocodes in this section provide a high-level summary of what

needs to be done rather than a technically working code example. Note that there may exist many

variants and implementations belonging to the same family of a layout algorithm.

A layout algorithm in essence takes in a graph G as a parameter and returns the positions of

all nodes. Since our output visualization is on the two-dimensional display medium, the algorithms

outlined below return the coordinate vectors x for all nodes. The supply network is represented

as a mathematical graph G. G is defined as a collection, (V,E), where V is the set of nodes (or

vertices) and E is the set of edges connecting the pairs of nodes.

Algorithm S1 shows a pseudocode for implementing the force layout. ↵ is the dampening

parameter over loops, which attenuates over time by the factor of k and determines when to

terminate the algorithm. The key in this algorithm is the computation of three forces applied to

each node in every iteration. The three forces are link-based elastic force (h), gravity (g), and

distance-based repulsion (f). Computation of these forces requires information about all node

positions (X) at the moment.

Algorithm S1 Force-Directed Layout

1: procedure Force(G := (V,E)) . Take in nodes V and edges E as parameters2: while ↵ > ↵ do . Repeat node place adjustment until convergence is reached3: for all v 2 V do . Loop over all nodes4: �xv ↵(hv(X) + gv(X) + fv(X)) . Compute forces and node displacement5: xv xv +�xv . Update position for node v6: end for7: ↵ k↵ . where k < 18: end while9: end procedure

S-2

Page 3: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

Algorithm S2 shows a pseudocode for implementing the chord layout. Nodes are represented

as arcs along a circle. The arc size is computed in proportion to the sum of indegree (deg�(v))

and outdegree (deg+(v)) of node v. Once all nodes are positioned, edges are placed so that each

edge connects a source node and a target node with a ribbon.

Algorithm S2 Chord Layout

1: procedure Chord(G := (V,E)) . Take in nodes V and edges E as parameters2: for i 1, . . . , |V | do . Loop over all nodes3: v Vi . v is the focal node in an iteration

4: lv deg�(v)+deg+(v)Pv2V (deg�(v)+deg+(v))

. Compute arc length for node v

5: ✓i,start ✓i�1,end . Compute angular position of each node v or Vi

6: ✓i,end ✓i,start + lv7: end for8: for all e 2 E do . Loop over all edges9: s, t source(e), target(e) . s, t are source and target of an edge e

10: Create a ribbon connection s and t where s 2 [✓s,start, ✓s,end] and t 2 [✓t,start, ✓t,end].11: end for12: end procedure

Algorithm S3 shows a pseudocode for implementing the treemap layout. Ideally, the treemap

assumes hierarchy in the data structure among nodes. Denote hierarchy levels from h = 1, . . . , H.

Each level of hierarchy consists a disjoint partition of V . Denote the partition at level h as Ph such

thatS

p2Php = V where p is a partition and a set of nodes. At the lowest level, individual nodes

become partitions.

Algorithm S3 Treemap Layout

1: procedure Treemap(G := (V,E)) . Take in nodes V and edges E as parameters2: for h = 1, . . . , H do . Loop from higher to lower hierarchy3: for all p 2 sorted(Ph) do . Loop from large to small partitions4: Compute the aggregate node attributes for all nodes in partition p.5: Assign rectangular area proportional to the aggregate measure.6: end for7: end for8: end procedure

S-3

Page 4: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

Algorithm S4 shows a pseudocode for implementing the matrix layout. The matrix layout is a

square grid that consists of square cells. Cells,t denotes row s and column t in the square grid.

Algorithm S4 Matrix Layout

1: procedure Matrix(G := (V,E)) . Take in nodes V and edges E as parameters2: Create a |V |⇥ |V | square grid.3: for all e 2 E do4: s, t source(e), target(e) . s, t are source and target of an edge e5: Color Cells,t with intensity based on edge attributes (e.g., weight).6: end for7: end procedure

S-4

Page 5: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

S.3. Supporting Tables about Data Schema and Visual Encodings

Table S1: Data Schema (Adapted from Willems [72])

Field Description

Stage Name Label for the stageStage Classification One of five types for the stage: Distribution, Manufacturing, Procurement, Retail, and TransportationStage Cost Amount of direct costs added at the stage denoted by dollarStage Time Processing time at the stage in daysDemand Daily demand rate at the stage reported only when the stage faces final demand

Table S2: List of Elements for Encoding Di↵erent Types of Variables

Type Encoding Elements Categorical Continuous

Node Size •Shape •Fill Color • •Fill Pattern •Border Line Color • •Border Line Thickness •Border Line Pattern •

Edge Line Color • •Line Thickness •Line Pattern •

Table S3: Available Visual Encodings for Di↵erent Layouts

Force-directed Circular Treemap Matrix

Node or Edge Attributes NS NC ET NS NC FS FC NS NC CC CI

Uniform • •Sum of Flows •Activity Type • • •Individual Stage Cost • • • •Downstream Cumulative Cost • • • •Average Stage Time • • • •In-degree • • • •Out-degree • • • •Out-degree-to-in-degree Ratio • • • •In-degree Centrality • • • •Out-degree Centrality • • • •Betweenness Centrality • • • •

Originating Stage’s Activity Type • •Originating Stage’s Individual Cost • • •Originating Stage’s Cumulative Cost • • •Originating Stage’s Time • • •Edge Betweenness Centrality •

Note: NS = Node Size, NC = Node Color, ET = Edge Thickness, FS = Flow Size,FC = Flow Color, CC = Cell Color, CI = Cell Intensity; Upper pane contains nodeattributes, while lower pane contains edge attributes.

S-5

Page 6: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

S.4. Histograms for User Study and Survey Results

(a) Evaluation results indicate positive value of the visual analyticssystem used for understanding supply networks. Scores are based onthe 5-point Likert scale. 1 is “Strongly Disagree” and 5 is “StronglyAgree”.

(b) Rank orders of preference show force-directed and substrate layouts stand out. Scores are based onrank order. 1 is highest and 5 is lowest.

Figure S1: Overall evaluation results prove positive value from the pilot study with MBA students.

S-6

Page 7: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

S.5. Large Size Images for Small Figures in Main Text

Figure S2: Large size image of Figure 3(a).

S-7

Page 8: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

Figure S3: Large size image of Figure 3(b).

S-8

Page 9: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

Figure S4: Large size image of Figure 3(c).

S-9

Page 10: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

Figure S5: Large size image of Figure 3(d).

S-10

Page 11: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

Figure S6: Large size image of Figure 4(a).

S-11

Page 12: Supplementary Material Data Publication Individual Data Items · S.3. Supporting Tables about Data Schema and Visual Encodings Table S1: Data Schema (Adapted from Willems [72]) Field

Figure S7: Large size image of Figure 4(b).

S-12