vault backup: 2025-03-16 18:59:42

AI & Data Mining/Week 5/Lecture 9 - PRISM.md

@@ -18,6 +18,7 @@
 ## Selecting a Test
 
 Goal: Maximise probability of desired class
+
 - $t$ = total number of examples covered by rule
 - $p$ = number of positive examples of the class covered by rule
 - $t - p$ = number of errors made by rule
@@ -44,7 +45,7 @@ Stop Condition: $t-p=0$
 
 ![](Pasted%20image%2020241017131912.png)
 
-#### Modified Rule and its Coverage
+#### Modified Rule and Its Coverage
 
 - Rule with best test added: If astigmatism = Yes and tear rate = Normal { then recommendation = Hard }
 ![](Pasted%20image%2020241017132019.png)
@@ -92,4 +93,4 @@ For each class C
 - Default Rule
 	- If no rules cover example, prediction is the majority class (most frequent in training data)
 - Conflict Resolution Strategy
-	- If more than one rule covers an example, select predicted class with highest recurrance in training data
+	- If more than one rule covers an example, select predicted class with highest recurrence in training data

AI & Data Mining/Week 5/Tutorial 9 - PRISM.md

@@ -41,6 +41,7 @@ IF Income = 0-15k THEN risk = high
 | debt = high              | 4/6     |
 | debt = low               | 2/4     |
 | **collateral = none**    | **6/6** |
+
 IF Income = 0-15k AND collateral = none THEN risk = high
 
 | Credit History | Debt | Collateral | Income | Risk |
@@ -59,5 +60,3 @@ IF Income = 0-15k AND collateral = none THEN risk = high
 | credit history = unknown | 1/4     |
 | debt = high              | 2/4     |
 | debt = low               | 2/4     |
-
-