Ai lessons

lessons/Data-Science/ai-decision-trees/langs/en-us/images/lesson-images.json

@@ -0,0 +1,68 @@
+{
+  "tree1.png" : {
+    "description" : "A portion of a decision tree (leading from utensil to knife) with two levels.",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+  "tree2.png" : {
+    "description" : "A portion of a decision tree (leading from utensil to knife, and from not-utensil to cup) with three levels.",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+  "tree3.png" : {
+    "description" : "A decision tree that categorizes knife, fork, spoon, plate, mug, and cup",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+  "incomplete-tree.png" : {
+    "description" : "A decision tree (with blanks) that categorizes knife, fork, spoon, plate, mug, and cup",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+  "tree4.png" : {
+    "description" : "A less-accurate decision tree that categorizes knife, fork, spoon, plate, mug, and cup.",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+  "age-stump.png" : {
+    "description" : "A one-level decision tree showing three branches (teens, twenties and thirties), coming from the root node of age.",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+   "age-stump-checks.png" : {
+    "description" : "A one-level decision tree showing three branches (teens, twenties and thirties), coming from the root node of age with checkmarks indicating correct predictions",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  }, 
+  "blank-stump.png" : {
+    "description" : "A blank decision tree stump",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+  "final-tree.png" : {
+    "description" : "A decision tree that predicts if individuals will buy a video game",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  }, 
+   "history-stump.png" : {
+    "description" : "A decision tree stump for the column shopping history",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+   "interest-stump.png" : {
+    "description" : "A decision tree stump for the column interest in game",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  }, 
+   "terminology-tree.png" : {
+    "description" : "A decision tree indicating where the root, nodes, and leaves are located",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+   "iris-data.png" : {
+    "description" : "An excerpt of Ronald Fisher's Iris dataset",
+    "source" : "Created by the Bootstrap",
+    "license" : "Creative Commons 4.0 - NC - SA",
+    "caption" : "British statistician and biologist Ronald Fisher first published his findings about the Iris Dataset in 1936. The dataset includes 150 different plants' sepal length, sepal width, petal length, and petal width. Today, this dataset is considered a go-to dataset useful for illustrating a wide range of problems in data science and machine learning."
+  }
+}

lessons/Data-Science/ai-decision-trees/langs/en-us/pages/build-and-test.adoc

@@ -0,0 +1,80 @@
+= Building and Testing a Decision Tree
+
+On @printable-exercise{level-1.adoc}, we decided that the decision mode at the end of the "teens" branch should be "interest in game". Let's create the rest of the tree and test it out. You can refer to the training dataset on @handout{decision-tree-data.adoc}.
+
+== Build and Understand the Tree
+
+@vspace{1ex}
+
+Fill in the remainder of the tree using the algorithm you have learned (left) before filling in the blanks (right).
+
+@vspace{1ex}
+
+[cols="1,1", stripes="none"]
+|===
+
+|
+@ifnotsoln{@center{@image{../images/incomplete-tree.png, 350}}}
+
+@ifsoln{@center{@image{../images/final-tree.png, 350}}}
+
+|
+
+@n The root node of this tree is @fitb{10em}{@ifsoln{age}}.
+
+
+@n The first set of branches includes @fitb{10em}{@ifsoln{teens}}, @fitb{10em}{@ifsoln{twenties}}, and @fitb{10em}{@ifsoln{thirties}}.
+
+@vspace{1ex}
+
+@n Write the rules that this decision tree follows: @ifsoln{Answers will vary.}
+
+
+a) Predict that @fitb{10em}{@ifsoln{interested teens}} will buy the game.
+
+@vspace{1ex}
+
+b) Predict that *all* individuals in their twenties @fitbruby{10em}{@ifsoln{will}}{will / will not} buy the game.
+
+c) Predict that @fitb{15em}{@ifsoln{new customers in their thirties}} will not buy the game.
+
+|===
+
+
+== Test the Tree
+
+@vspace{1ex}
+
+Let's see how well our decision tree predicts if Kat, Billy, or Chen will buy the game.
+
+@vspace{1ex}
+
+[cols="2,2,3,2,2", stripes="none", options="header"]
+|===
+
+| name 		| age 		| shopping history 	| interest in game 	| buys game
+| Kat 		| teen 		| new customer		| yes 				| yes
+| Billy		| twenties	| new customer		| no 				| no
+| Chen	    | twenties 	| previous customer | no  				| no
+
+|===
+
+@n What prediction will the decision tree make for Kat? @fitb{10em}{@ifsoln{will buy}} Billy? @fitb{10em}{@ifsoln{will buy}} Chen? @fitb{10em}{@ifsoln{will buy}}
+
+@n Compare the test data with the decision tree recommendations. How often was the computer correct? @fitb{}{@ifsoln{1 out of 3 times}}
+
+@n Based on this new test data, which rule (see Q3, above) needs to change? Why? @fitb{}{@ifsoln{We need to change Rule b,}}
+
+@fitb{}{@ifsoln{that all individuals in their 20s will buy the game. The new test data contradicts the rule.}}
+
+== Reflect
+
+@n The decision tree we built was correct 100% of the time with our training dataset, but correct only 33% of with our test dataset. Why?
+
+@fitb{}{@ifsoln{The training dataset is small, and only represents a sample of the population. The sample was not representative,}}
+
+@fitb{}{@ifsoln{as there are individuals in their twenties who choose not to purchase, among others who contradict the training data.}}
+
+@n What could we do to improve the quality of this decision tree? @fitb{}{@ifsoln{We would need to collect a bigger, more representative dataset,}}
+
+@fitb{}{@ifsoln{and then build the tree again.}}

lessons/Data-Science/ai-decision-trees/langs/en-us/pages/comparing-trees.adoc

@@ -0,0 +1,60 @@
+= Comparing Decision Trees
+
+Use the two decision trees (below) to respond to the prompts.
+
+[cols="^.^1,^.^1", stripes="none"]
+|===
+
+|
+
+@image{../images/tree3.png, 300}
+
+|
+
+@image{../images/tree4.png, 290}
+
+|===
+
+@n What do you Notice about the two decision trees above?
+
+@fitb{}{@ifsoln{The trees are different because the "not flat" branch of the tree on the right terminates with "mug or cup".}}
+
+@fitb{}{}
+
+@fitb{}{}
+
+@n What do you Wonder about the two decision trees?
+
+@fitb{}{}
+
+@fitb{}{}
+
+@fitb{}{}
+
+@n The tree on the @fitbruby{10em}{@ifsoln{left}}{[left / right]} is more accurate, while the tree on the @fitbruby{10em}{@ifsoln{right}}{[left / right]} is more efficient.
+
+@n With AI, there is often a trade-off between *efficiency* and *accuracy*. Explain what this statement means to you by describing the efficiency and accuracy of the two example trees.
+
+@fitb{}{@ifsoln{The tree with four nodes cannot label the six listed items with 100% accuracy. It cannot distinguish between "cup" and "mug".}}
+
+@fitb{}{@ifsoln{The tree with five nodes is more accurate and knows the difference between "cup" and "mug", but it is less efficient.}}
+
+@fitb{}{}
+
+@n How will the decision tree on the left label the items below? Explain your response.
+
+@indented{
+a) knife @fitb{}{@ifsoln{"knife"}}
+
+@fitb{}{}
+
+b) chopstick @fitb{}{@ifsoln{uncertain response... maybe "mug"?}}
+
+@fitb{}{}
+
+c) spork @fitb{}{@ifsoln{"fork"}}
+
+@fitb{}{}
+
+}
+

lessons/Data-Science/ai-decision-trees/langs/en-us/pages/decision-tree-data.adoc

@@ -0,0 +1,25 @@
+= Decision Tree Training Dataset
+
+
+
+[cols="2,2,3,2,2", stripes="none", options="header"]
+|===
+
+| name 		| age| shopping history 	|  interest in game | buys game
+
+| Jan 		| 16 | previous customer 	| no  				| no
+| Jose 		| 19 | previous customer 	| no 				| no
+| Maribel 	| 21 | previous customer  	| no  				| yes
+| Noah		| 33 | previous customer	| no  				| yes
+| Sydney 	| 36 | previous customer 	| yes 				| yes
+| Mariana 	| 32 | new customer 		| yes 				| no
+| Rasula	| 24 | new customer 		| yes 				| yes
+| Jillian	| 14 | previous customer 	| no 				| no
+| Ariella	| 16 | new customer  		| yes  				| yes
+| Isabela	| 38 | previous customer	| yes 				| yes
+| Danial	| 19 | previous customer 	| yes 				| yes
+| Kate		| 27 | previous customer	| no 				| yes
+| Taikhoom	| 22 | previous customer 	| yes  				| yes
+| Peter 	| 35 | new customers		| no 				| no
+
+|===

lessons/Data-Science/ai-decision-trees/langs/en-us/pages/level-1.adoc

@@ -0,0 +1,53 @@
+= Decision Tree: Level 1
+
+[.linkInstructions]#Use the @handout{decision-tree-data.adoc, "training data"} to respond to the prompts below.# Let's investigate what happens when we begin our tree with "age" as the root.
+
+== Investigating "age"
+
+@n Using information from the "age" column and the "buys game" column, fill in the blanks below.
+
+- The training data indicates that individuals in their thirties probably @fitbruby{10em}{@ifsoln{will}}{will / will not} buy the game.
+
+- The training data indicates that individuals in their twenties probably @fitbruby{10em}{@ifsoln{will}}{will / will not} buy the game.
+
+- The training data indicates that teens @fitbruby{10em}{@ifsoln{will not}}{will / will not} buy the game.
+
+
+Here’s our rule: 
+
+- *Predict that individuals in their @fitbruby{15em}{@ifsoln{twenties}}{teens / twenties / thirties} and in their @fitbruby{15em}{@ifsoln{thirties}}{teens / twenties / thirties} will buy the game.* 
+- *Predict that individuals in their @fitbruby{15em}{@ifsoln{teens}}{teens / twenties / thirties} will not.*
+
+@n Will a computer following this rule make correct predictions for all 14 individuals in the training dataset? @fitb{5em}{@ifsoln{No.}}
+
+@n According to our training data, this rule is correct in @fitb{5em}{@ifsoln{10}} instances out of 14.
+
+@n Here is one way to represent the decision tree for "age". We call this a "decision stump" because it has just one level. Place checkmarks below all correctly predicted values in the "twenties" and "thirties" leaf nodes.
+
+@ifnotsoln{@center{@image{../images/age-stump.png, 250}}}
+
+@ifsoln{@center{@image{../images/age-stump-checks.png, 250}}}
+
+@n Are any of the leaf nodes for "age" pure? (_Pure_ means that the leaf contains exclusively one single value.) @fitb{5em}{@ifsoln{Yes; "twenties".}}
+
+@n If we used the decision tree stump above to make predictions:
+
+- the likelihood of a correct prediction for a teenage shopper is @fitb{5em}{@ifsoln{60}}%
+
+- the likelihood of a correct prediction for a shopper in their twenties is @fitb{5em}{@ifsoln{100} }%
+
+- the likelihood of a correct prediction for a shopper in their thirties? @fitb{5em}{@ifsoln{60} }%
+
+@vspace{1ex}
+
+== Branching beyond "age"
+
+@n After creating the first level of his decision tree, Ernie removed Maribel, Rasula, Kate, and Taikhoom from the training set. As Bert watched Ernie grab a pencil and cross out those rows, he said, "I don't think that's a good idea! Why would we alter our dataset just because we're starting the second level of the tree?" Why does it make sense to remove those four individuals from the dataset? Explain Ernie's (correct) decision to Bert.
+
+@fitb{}{}
+
+@fitb{}{}
+
+@n For each decision node we have created ("teens" and "thirties"), we need to decide whether it makes more sense to branch to @fitbruby{15em}{@ifsoln{interest in game}}{[column]} or @fitbruby{15em}{@ifsoln{shopping history}}{[column]}. We do not need to branch from "twenties" because @fitb{}{}
+
+@fitb{}{}

lessons/Data-Science/ai-decision-trees/langs/en-us/pages/level-2.adoc

@@ -0,0 +1,60 @@
+= Decision Tree: Level 2
+
+[.linkInstructions]#Use the @handout{decision-tree-data.adoc, "training data"} to respond to the prompts below.# Let's decide what decision nodes to use in the 2nd level of our tree.
+
+== Shopping History as the Decision Node
+
+@n Using information from the "age" column and the "shopping history" column, fill in the blanks below.
+
+@indented{a) The training data indicates that *frequent customers in their teens* @fitbruby{10em}{@ifsoln{will}}{will / will not} buy the game.
+
+b) The training data indicates that *new customers in their teens* probably @fitbruby{10em}{@ifsoln{will}}{will / will not} buy the game.
+
+@vspace{1ex}
+
+c) What does the training data on *infrequent customers in their teens* tell us? @fitb{}{}
+
+@fitb{}{@ifsoln{The data does not provide any information on the likelihood of game purchase.}}
+}
+
+
+@n Rule: *Predict that frequent teenager shoppers @fitbruby{5em}{@ifsoln{will}}{will / will not} buy the game and that new teenage customers @fitbruby{5em}{@ifsoln{will}}{will / will not} buy the game.*
+
+@n Will a computer following this rule make the correct prediction every time? @fitb{5em}{@ifsoln{No.}}
+
+@n According to our training data, this rule is correct in @fitb{5m}{@ifsoln{four}} instances out of five.
+
+@n Create a decision stump for "shopping history" with a root node of "teens". Place checkmarks below all correctly predicted values in the leaf node.
+
+@ifnotsoln{@center{@image{../images/blank-stump.png, 150}}}
+
+
+@ifsoln{@center{@image{../images/history-stump.png, 150}}}
+
+
+== Interest in Game as the Decision Node
+
+@n Using information from the "age" column and the "interest in game" column, fill in the blanks below.
+
+@indented{
+
+a) The training data indicates that *interested teens*  @fitbruby{10em}{@ifsoln{will}}{will / will not} buy the game.
+
+b) The training data indicates *uninterested teens* probably @fitbruby{10em}{@ifsoln{will}}{will / will not} buy the game.
+}
+
+@n Rule: *Predict that interested teenager shoppers @fitbruby{5em}{@ifsoln{will}}{will / will not} buy the game and uninterested teenagers shoppers @fitbruby{5em}{@ifsoln{will}}{will / will not}.*
+
+@n Will a computer following this rule make the correct prediction every time? @fitb{5em}{@ifsoln{Yes!}}
+
+@n Create a decision stump for teenagers' interest in the game.Place checkmarks below all correctly predicted values in the leaf node.
+
+@ifnotsoln{@center{@image{../images/blank-stump.png, 150}}}
+
+@ifsoln{@center{@image{../images/interest-stump.png, 150}}}
+
+== What decision attribute should we use?
+
+@n Shopping history or interest? Explain your response. @fitb{}{@ifsoln{interest in game}}
+
+@fitb{}{}

lessons/Data-Science/ai-decision-trees/langs/en-us/pages/workbook-pages.txt

@@ -0,0 +1,4 @@
+level-1.adoc
+level-2.adoc
+build-and-test.adoc
+comparing-trees.adoc

lessons/Data-Science/ai-decision-trees/langs/en-us/proglang.txt

@@ -0,0 +1 @@
+pyret

lessons/Data-Science/ai-decision-trees/langs/en-us/slides-pyret.id

@@ -0,0 +1 @@
+

lessons/Data-Science/ai-linear-regression/langs/en-us/images/lesson-images.json

@@ -0,0 +1,22 @@
+{
+  "navlab.png" : {
+    "description" : "ALVINN / the Navlab, a retrofitted self-driving army ambulance",
+    "source" : "Carnegie Mellon Robotics Institute, https://www.ri.cmu.edu/robot/navlab-ii/",
+    "license" : "Creative Commons 4.0 - NC - SA"
+  },
+  "road1.png" : {
+    "description" : "US Route 191 near Monomunt Valley",
+    "source" : "Wikimedia Commons",
+    "license" : "Creative Commons Attribution-Share Alike 2.0 Generic"
+  },
+  "road2.png" : {
+    "description" : "curved concrete road surrounded with trees",
+    "source" : "picpik.com",
+    "license" : "Creative Commons 1.0 Public Domain"
+  },
+  "road3.png" : {
+    "description" : "curved road in South Moravia",
+    "source" : "rawpixel.com",
+    "license" : "CCO 1.0 Universal"
+  }
+}