Machine Learning with

About Bryan Cafferky...

alt text

email: bryan256.@msn.com

blog: sql-fy.com

Twitter: @BryanCafferky

LinkedIn: linkedin.com/in/bryancafferky

Experience

Author: Pro PowerShell for Database Developers by Apress.
Found and Lead The RI MS BI User Group.
Health Care, Banking, Insurance, ecommerce, utilities.
Frequent technical speaker at technical events.
Decades of IT experience.

Presentation Goals

Demonstrate how to apply Machine Learing with R

While Using R's Dynamic Documents features to show the usefulness of these and how they are applied.

Publishing Dynamic Documents in R

Problem:

With technical presentations, you need to cut and paste code samples into slides and switch between the slides and the IDE.

The goal of dynamic documents is to maintain the code with the presentation.

Saves time and effort!

Publishing Dynamic Documents in R

Markdown (books, slides)
Beamer (slides)
LaTeX
R HTML
Shiny (interactive web apps, think RSP)
FlexDashboard (interactive web apps)
knitr integrates some of the above technologies.

Machine Learning Steps

Profile the Data
Determine the columns to be used for input and what is to be predicted.
Determine what model(s) fit the data best.
Train the Model - Example
Evaluate the Model
Determine the Best Model
Use the Model

Machine Learning - The Matrix of Stats

Machine Learning - Profiling the Data

Str (structure of the data)
Summary (basic stats)
Head/Tail (view the data)
Scatter Plot (what's the distribution)
Histogram (normal distribution?)
Box Plot (visual summary)

Machine Learning - Statistics

Central Tendency
Range of Values
Distibution
Relationships

Machine Learning - Profiling Data - str

str will return the structure of the data.

str(iris, vec.len=1)

'data.frame':   150 obs. of  5 variables:
 $ Sepal.Length: num  5.1 4.9 ...
 $ Sepal.Width : num  3.5 3 ...
 $ Petal.Length: num  1.4 1.4 ...
 $ Petal.Width : num  0.2 0.2 ...
 $ Species     : Factor w/ 3 levels "setosa","versicolor",..: 1 1 ...

Machine Learning - Profiling the Data - Summary

See the basic statistics…

summary(iris)

Sepal.Length	Sepal.Width	Petal.Length	Petal.Width	Species
Min. :4.300	Min. :2.000	Min. :1.000	Min. :0.100	setosa :50
1st Qu.:5.100	1st Qu.:2.800	1st Qu.:1.600	1st Qu.:0.300	versicolor:50
Median :5.800	Median :3.000	Median :4.350	Median :1.300	virginica :50
Mean :5.843	Mean :3.057	Mean :3.758	Mean :1.199	NA
3rd Qu.:6.400	3rd Qu.:3.300	3rd Qu.:5.100	3rd Qu.:1.800	NA
Max. :7.900	Max. :4.400	Max. :6.900	Max. :2.500	NA

Machine Learning - Profiling the Data - Viewing the Data - Head

See the first few rows…

head(iris)

Sepal.Length	Sepal.Width	Petal.Length	Petal.Width	Species
5.1	3.5	1.4	0.2	setosa
4.9	3.0	1.4	0.2	setosa
4.7	3.2	1.3	0.2	setosa
4.6	3.1	1.5	0.2	setosa
5.0	3.6	1.4	0.2	setosa
5.4	3.9	1.7	0.4	setosa

Machine Learning - Profiling the Data - Viewing the Data - Tail

See the last few rows…

tail(iris)

	Sepal.Length	Sepal.Width	Petal.Length	Petal.Width	Species
145	6.7	3.3	5.7	2.5	virginica
146	6.7	3.0	5.2	2.3	virginica
147	6.3	2.5	5.0	1.9	virginica
148	6.5	3.0	5.2	2.0	virginica
149	6.2	3.4	5.4	2.3	virginica
150	5.9	3.0	5.1	1.8	virginica

Machine Learning - Scatterplot

See Clustering and Distribution

plot(iris$Petal.Width,pch=16)

plot of chunk plot

Machine Learning - Histogram

Does the data fit a normal curve?

hist(iris$Petal.Width,x.lab="petal width", col="lightblue")

plot of chunk unnamed-chunk-8

Machine Learning - Quartiles

Machine Learning - Box Plot Explained

Machine Learning - Box Plot

Visualize the Summary Statistics

boxplot(iris,col=c("blue","red","sienna","palevioletred1","blue"),ylab="Centimeters")

plot of chunk unnamed-chunk-9

Machine Learning - Born Statistician

Machine Learning - Determine Columns

Ockham's Razor - Minimal columns is best…

myiris <- 
  data.frame(iris$Sepal.Length,iris$Sepal.Width,iris$Petal.Length,iris$Petal.Width)

d <- data.frame(myiris)
dcor <- as.matrix(cor(d))
dcor

	iris.Sepal.Length	iris.Sepal.Width	iris.Petal.Length	iris.Petal.Width
iris.Sepal.Length	1.0000000	-0.1175698	0.8717538	0.8179411
iris.Sepal.Width	-0.1175698	1.0000000	-0.4284401	-0.3661259
iris.Petal.Length	0.8717538	-0.4284401	1.0000000	0.9628654
iris.Petal.Width	0.8179411	-0.3661259	0.9628654	1.0000000

Machine Learning - Determine Columns

library(psych)
pairs.panels(iris)

Machine Learning - Social Media Humor

Machine Learning - Normal Distribution

Machine Learning - Which model to use?

What is the answer?

Machine Learning - Which model to use?

What is the answer?

Machine Learning - Which model to use?

What is the answer?

Machine Learning - Which model to use?

What is the answer?

What is the question?

Machine Learning - Determining the Model

Distribution
Type of Data
- True/False
- Number
- Infinite (continuous)
- Finite List
- Text (Category)
How many variables (parameters/features) determine result
Type of relationship between parameters and result

Machine Learning - Azure ML Cheat Sheet

$alt text$

Machine Learning - Enterprise Statistics

Machine Learning - Iris Data Set

Train the Model: Separate Training/Test Data

#  Separate my training and scoring data sets...
set.seed(1234)
ind <- sample(2, nrow(iris), replace=TRUE, prob=c(0.67, 0.33))

iris.training <- iris[ind==1, 1:4]
iris.test <- iris[ind==2, 1:4]

#  What we want to predict...
iris.trainLabels <- iris[ind==1, 5]
iris.testLabels <- iris[ind==2, 5]

Machine Learning - How KNN Works

k = number of closest neighbors to consider
Each color = different species

Train the model: knn (k-Nearest Neighbors)

library(class)
iris_pred <- knn(train = iris.training, test = iris.test, cl = iris.trainLabels, k=3)
iris_pred

 [1] setosa     setosa     setosa     setosa     setosa     setosa    
 [7] setosa     setosa     setosa     setosa     setosa     setosa    
[13] versicolor versicolor versicolor versicolor versicolor versicolor
[19] versicolor versicolor versicolor versicolor versicolor versicolor
[25] virginica  virginica  virginica  virginica  versicolor virginica 
[31] virginica  virginica  virginica  virginica  virginica  virginica 
[37] virginica  virginica  virginica  virginica 
Levels: setosa versicolor virginica

Evaluate the model: knn (k-Nearest Neighbors)

#  Score the model...
library(gmodels)

CrossTable(x = iris.testLabels, y = iris_pred, prop.chisq=FALSE)

No model is saved.
Test data set is what you want predicted.
Model must be regenerated each time.
We manually evaluate effectiveness.

Evaluate the model: knn Output

An Alternative Mode: Decision Tree

# From  http://davetang.org/muse/2013/03/12/building-a-classification-tree-in-r/

 library(tree)

 tree1 <- tree(Species ~ Sepal.Width + Sepal.Length + Petal.Length + Petal.Width,data = iris)

 plot(tree1)
 text(tree1)

An Alternative Mode: Decision Tree Output

plot of chunk unnamed-chunk-17

An Alternative Mode: Decision Tree: Usage

  # Use th model to get classification...
 predict.list <- predict(tree1, iris.test, type="class")
 predict.list

 [1] setosa     setosa     setosa     setosa     setosa     setosa    
 [7] setosa     setosa     setosa     setosa     setosa     setosa    
[13] versicolor versicolor versicolor versicolor versicolor versicolor
[19] versicolor versicolor versicolor versicolor versicolor versicolor
[25] virginica  virginica  virginica  virginica  virginica  virginica 
[31] virginica  virginica  virginica  virginica  virginica  virginica 
[37] virginica  virginica  virginica  virginica 
Levels: setosa versicolor virginica

Evaluate the model: knn (k-Nearest Neighbors)

#  Score the model...
library(gmodels)
CrossTable(x = iris.testLabels, y =  predict.list, prop.chisq=FALSE)

Evaluate the model: knn (k-Nearest Neighbors)

Machine Learning - Born Statistician

Review: Machine Learning Steps

Profile the Data
Determine the columns to be used for input and what is to be predicted.
Determine what model(s) fit the data best.
Train the Model - Example
Evaluate the Model
Determine the Best Model
Use the Model

Machine Learning with

About Bryan Cafferky...

Presentation Goals

Demonstrate how to apply Machine Learing with R

While Using R's Dynamic Documents features to show the usefulness of these and how they are applied.

Publishing Dynamic Documents in R

Problem:

With technical presentations, you need to cut and paste code samples into slides and switch between the slides and the IDE.

The goal of dynamic documents is to maintain the code with the presentation.

Saves time and effort!

Publishing Dynamic Documents in R

Machine Learning Steps

Machine Learning - The Matrix of Stats

Machine Learning - Profiling the Data

Machine Learning - Statistics

Machine Learning - Profiling Data - str

str will return the structure of the data.

Machine Learning - Profiling the Data - Summary

See the basic statistics…

Machine Learning - Profiling the Data - Viewing the Data - Head

See the first few rows…

Machine Learning - Profiling the Data - Viewing the Data - Tail

See the last few rows…

Machine Learning - Scatterplot

See Clustering and Distribution

Machine Learning - Histogram

Does the data fit a normal curve?

Machine Learning - Quartiles

Machine Learning - Box Plot Explained

Machine Learning - Box Plot

Visualize the Summary Statistics

Machine Learning - Born Statistician

Machine Learning - Determine Columns

Ockham's Razor - Minimal columns is best…

Machine Learning - Determine Columns

Machine Learning - Social Media Humor

Machine Learning - Normal Distribution

Machine Learning - Which model to use?

What is the answer?

Machine Learning - Which model to use?

What is the answer?

Machine Learning - Which model to use?

What is the answer?

Machine Learning - Which model to use?

What is the answer?

What is the question?

Machine Learning - Determining the Model

Machine Learning - Azure ML Cheat Sheet

Machine Learning - Enterprise Statistics

Machine Learning - Iris Data Set

Train the Model: Separate Training/Test Data

Machine Learning - How KNN Works

Train the model: knn (k-Nearest Neighbors)

Evaluate the model: knn (k-Nearest Neighbors)

Evaluate the model: knn Output

An Alternative Mode: Decision Tree

An Alternative Mode: Decision Tree Output

An Alternative Mode: Decision Tree: Usage

Evaluate the model: knn (k-Nearest Neighbors)

Evaluate the model: knn (k-Nearest Neighbors)

Machine Learning - Born Statistician

Review: Machine Learning Steps

Final Review

Demonstrated how to apply Machine Learing with R

While Using R's Dynamic Documents features to show the usefulness of these and how they are applied.

Q & A