Skip to content

Latest commit

 

History

History
78 lines (52 loc) · 3.24 KB

methodology.md

File metadata and controls

78 lines (52 loc) · 3.24 KB

Methodology

Figure 1: Research methodology of the study

Data

Data transformation

To generate a network of rules and samples, following are the steps:

  1. Conversion of the decision tree rules and samples to a network
  2. Network analysis

Conversion of the decision tree to network

The following steps are implemented for transformation of the dataset.

  1. Conversion of decision tree rules from for each sample
    • Serialization: Converting decision tree rules to csv files for each sample.

🧪 Formulation of the rule and sample network

The following is the formulation of the idea (rule and sample network) using definition / notation from graph theory.

Mathematical definition and formulation

The following is the mathematical formualtion of the rules and sample network Formally, a graph is sets $(V, E)$, where [1]:

  • $V$ is a non-empty set whose elements are called $vertices$, each element in $V$ is a sample, or a point of data in the given dataset.
  • $E$ is a collection of two-element subsets of $v \in V$ called $Edges$. Each edge $e \in E$ denotes a rule association of two-element subset.

The network of the $E$ and $V$ is defined as: $$ N = (V, E) $$

We use $A--B$ to denote an edge between vertices $A$ and $B$ rather than the set notation ${A, B}$. Note that $A--B$ and $B--A$ are the same edge, just as ${A, B}$ and ${B, A}$ are the same set.

TODO

  • How to define a cluster?
  • How to define graphical properties of rules and samples?
    • What is true for a specific link and node?
  • How to describe that there is a value to the link as well?
  • How to describe that there is a value to the size of node as well?

Edge cases

  • The overall graph will be a disconnected graph, where there does not exist any path between at least one pair or vertices is called as a disconnected graph.
  1. There will be two clusters: i
  2. Every node should be connected to at least with a link node. Because, every node comes from root node.
  3. If there is a commonality between the nodes then there is a link.
  4. There can not be duplicate links in a system.

Similar techniques

Conceptual graph analysis

  • Conceptual graph analysis was developed by Grasser and Murachver in 1985 to get detailed knowledge from computer science experts and found away of representing it in a coherent fashion. There was a transformation of nodes an questions of the original method and have extended its application from information design to instructional design.

  • Conceptual graph analysis are fundamentally different from rule and graph objects.

Assumptions

  1. Expert knowledge can be gained through structured and unstructured interview.
  2. Expert knowledge can be graphed and labeled in a graph

Steps

  1. Clarify the uses of the graph information
  2. Choose a set of situations for the expert to analyze
  3. Construct a rough graph
  4. Prepare a list of follow-up questions
  5. Expand the graph
  6. Review the final graph

References

  1. Conceptual Graph Analysis - a brief introduction