1. What is a network?
A network is a set of nodes interconnected [1] by their edges. In dynamic data structures there are also pointers that determine the direction towards which nodes interact. In our case, in the sourcing environment the buyer points towards the supplier .
If we wanted to visualize both the suppliers and buyers that are part of the Suplos network , we would see something similar to the previous graph. As a centralizer and facilitator of the network, Suplos is at the core, with strong links that, if we wanted to understand better, we must explore the network’s own characteristics in more detail. ( At the end of this post you can see an image of the organization of the Suplos network ).
2. Alpha-Centrality
Also known as Katz centrality, it is a measure of centrality of a node within a network. Firstly, the closer a node (suppliers) is to internal sources, the more important and positive influence that node exerts on the interaction of the neighbors that connect it (buyers) .
Similarly, the more central a node (suppliers) is from external sources , the stronger and more positive its influence is on the network (Suplos ecosystem) .
It is possible to calculate through power processes the amount of influence that some suppliers exert on some buyers. If we organized the information by alpha-centrality in descending order, we would see the following from the five most relevant nodes:
| Alpha-Centrality | |
|---|---|
| 1. Nodo 900.06… | 518 |
| 2. Nodo: 900.53… | 323 |
| 3. Nodo: 860.53… | 216 |
| 4. Nodo: 811.03… | 201 |
| 5. Nodo: 900.16… | 201 |
3. Page-Rank
It is perhaps one of the most used algorithms in the world, it was registered in 1999 and forms the basis of the recommendation engine that Google uses to order its searches.
PageRank numerically assigns the relevance of nodes indexed by a search engine . Each link that a node receives counts as a vote, but the relevance of the node that casts the vote is also computed within the calculation. The quality of the node and the number of links received determine the influence ranking.
The previous graph shows us that there are more than 50 nodes with the capacity to influence and they are considered by the algorithm as relevant to the network. These nodes are not only represented by buyers, that is, some suppliers are decisive in the stabilization of the network.
4. In-Degree
In Analytics the degree of a node in a network (sometimes incorrectly referred to as connectivity) is the number of connections or edges a node has with others. If a network is directed, meaning that the edges point in one direction from one node to another node, then the nodes have two different degrees, the In -degree , which is the number of incoming edges. , and the Out-degree , which is the number of outgoing edges[2].
It is possible to calculate a degree probability function of a network, defined as P( k )= n k / n , where, n k is the number of nodes that have k degrees and n is the total number of nodes.
For our exercise, we focus on obtaining the numerical results of the entry degrees, that is, on those nodes that receive the award of Goods and/or Services.
| In-Degree | |
|---|---|
| 1. Nodo 900.06… | 517 |
| 2. Nodo: 900.53… | 322 |
| 3. Nodo: 860.53… | 215 |
| 4. Nodo: 811.03… | 200 |
| 5. Nodo: 900.16… | 200 |
When comparing the two previous algorithms we found that they have a very positive level of correlation[3], obtaining almost identical results for the 5 most influential nodes.
5. Suplos Network
Take aways
- The Network shows the events awarded in Suplos, evidencing the cardinality of the buyer -> supplier pointers.
- More than 10 demand centralizing nodes are evident, with 3 nodes especially relevant.
- In relation to the interaction between demand aggregators, their proximity is associated either by their geographical location or by belonging to the same economic sector.
- The closer a node (supplier) is to its core, the stronger the link , and therefore its influence towards its buyer.
- There are centralizing nodes that appear as “islands”, this is because they are buyers with very recent interactions or because their geographical location or specialization of their economic sector does not allow suppliers to be shared within the network
- The results of the influence and relevance rankings (PageRank, In-Degree, Alpha-Centrality) of nodes that offer goods or services are closer to the center of the network and links close to their centralizing node.
[1] Wikipedia
[2] Wikipedia
[3] Se transforma en logaritmo para realizar un diagnóstico más evidente de la relación entre los algoritmos.
