#	Table of Content
1	Classification and Regression Tree (CART) Theory
1.1	History of CART
1.2	Machine Learning (CART)
1.3	Classification versus Regression
1.4	Umbrella of CART
1.5	Roots of our LNOB trees
2	ESCAP LNOB Trees
2.1	LNOB tree specifications
2.2	Indicators, Response Variables
2.3	Circumstances, Independent Variables
2.4	Source of data
2.5	LNOB Tree Example
2.5.1	LNOB Branches
2.5.2	LNOB Nodes
2.5.3	Furthest ahead group
2.5.4	LNOB Terminal nodes
2.5.5	Furthest behind group
2.6	Feature Space Diagram
2.6.1	All data points
2.6.2	1st split
2.6.3	2nd split
2.6.4	3rd split
2.7	LNOB Trees vs. Feature Space Diagram
2.8	Furthest behind group in a Feature Space Diagram
2.9	Furthest behind group in an LNOB Tree
2.10	Pros and cons of LNOB Trees
3	Dissimilarity Index
3.1	Origins of D-Index
3.2	D-index Scale
3.3	Video on D-Index
3.4	D-Index formula
3.4.1	Number of groups
3.4.2	Calculating number of groups
3.4.3	Country average access rate
3.4.4	Group access rate
3.4.5	Weighted sampling of group
3.4.6	Example of calculating D-Index
3.5	D-Index and barriers

2nd split

Next we can split the data in three different ways:

Lower + secondary grouped together, Higher
Lower, secondary + higher grouped together
Lower, Secondary, Higher all independently
Our algorithm checked all 3 possibilities and found the highest variance in the second combination. Lower as one group with an average access rate of 67% and secondary+higher as a second group with an average access rate of 88%

Introductory training