Prediction

Libraries and Datasets



In [135]:

    
source("https://raw.githubusercontent.com/eogasawara/mylibrary/master/myPreprocessing.R")
source("https://raw.githubusercontent.com/eogasawara/mylibrary/master/myPrediction.R")



In [136]:

    
load(url("https://github.com/eogasawara/mylibrary/raw/master/data/iris.RData"))
head(iris)









    





A data.frame: 6 × 5

	Sepal.Length Sepal.Width Petal.Length Petal.Width Species
	<dbl> <dbl> <dbl> <dbl> <fct>


	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

Building samples (training and testing)



In [137]:

    
iris_tt = sample.random(iris)
iris_train = iris_tt[[1]]
iris_test = iris_tt[[2]]

tbl <- rbind(table(iris$Species), table(iris_train$Species), table(iris_test$Species))
rownames(tbl) <- c("dataset", "training", "test")
head(tbl)









    





A matrix: 3 × 3 of type int

	 setosa versicolor virginica


	dataset 50 50 50
	training 42 37 41
	test  8 13  9

Majority class baseline prediction (Zero Rule)

Model creating and level of adjustment during training



In [138]:

    
model <- ZeroRule_model <- class_ZeroRule(iris_train, "Species")
head(model$train$metrics)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.5666667 0.35 0.35 0.675 0.35 0.35

Prediction using Zero Rule



In [139]:

    
test <- ZeroRule_test <- class_test(model, iris_test, "Species")
head(test$predictions)









    





A matrix: 6 × 3 of type dbl

	1 0 0
	1 0 0
	1 0 0
	1 0 0
	1 0 0
	1 0 0

Quality of prediction using Zero Rule



In [140]:

    
head(test$metrics)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.5111111 0.2666667 0.2666667 0.6333333 0.2666667 0.2666667

Confusion matrix



In [141]:

    
head(test$conf_mat)









    





       predictions
targets  1
      1  8
      2 13
      3  9

Decision Tree

Training the model, presenting the level of adjustment, quality of prediction, and confusion matrix.



In [142]:

    
model <- tree_model <- class_tree(iris_train, "Species")
head(model$train$metrics)
test <- class_test(model, iris_test, "Species")
head(test$metrics)
head(test$conf_mat)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9888889 0.9833333 0.9833333 0.9916667 0.9833333 0.9833333









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9111111 0.8666667 0.8666667 0.9333333 0.8666667 0.8666667









    





       predictions
targets  1  2  3
      1  8  0  0
      2  0 11  2
      3  0  2  7

Understing the produced model. Take a time comparing with the exploratory analysis.



In [143]:

    
plot_size(5, 4)
plot(tree_model$model)
text(tree_model$model)

Naive Bayes



In [144]:

    
model <- nb_model <- class_naiveBayes(iris_train, "Species")
head(model$train$metrics)
test <- class_test(model, iris_test, "Species")
head(test$metrics)
head(test$conf_mat)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9888889 0.9833333 0.9833333 0.9916667 0.9833333 0.9833333









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.8888889 0.8333333 0.8333333 0.9166667 0.8333333 0.8333333









    





       predictions
targets  1  2  3
      1  8  0  0
      2  0 10  3
      3  0  2  7



In [145]:

    
print(nb_model$model)









    



Naive Bayes Classifier for Discrete Predictors

Call:
naiveBayes.default(x = X, y = Y, laplace = laplace)

A-priori probabilities:
Y
    setosa versicolor  virginica 
 0.3500000  0.3083333  0.3416667 

Conditional probabilities:
            Sepal.Length
Y                [,1]      [,2]
  setosa     5.007143 0.3557149
  versicolor 5.945946 0.5118564
  virginica  6.602439 0.5511297

            Sepal.Width
Y                [,1]      [,2]
  setosa     3.454762 0.3603537
  versicolor 2.735135 0.3284782
  virginica  2.990244 0.3104874

            Petal.Length
Y                [,1]      [,2]
  setosa     1.461905 0.1820646
  versicolor 4.232432 0.4761268
  virginica  5.524390 0.5146749

            Petal.Width
Y                 [,1]      [,2]
  setosa     0.2452381 0.1130560
  versicolor 1.3000000 0.1885618
  virginica  2.0292683 0.2431501

Random Forest



In [146]:

    
model <- rf_model <- class_randomForest(iris_train, "Species")
head(model$train$metrics)
test <- class_test(model, iris_test, "Species")
head(test$metrics)
head(test$conf_mat)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	1 1 1 1 1 1









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9111111 0.8666667 0.8666667 0.9333333 0.8666667 0.8666667









    





       predictions
targets  1  2  3
      1  8  0  0
      2  0 11  2
      3  0  2  7

Prepare dataset for Machine Learning



In [147]:

    
iris_train_n <- normalize.minmax(iris_train)
iris_train_n$data$Class <- iris_train$Class
iris_test_n  <- normalize.minmax(iris_test, iris_train_n$norm.set)
iris_test_n$data$Class <- iris_test$Class

Neural Networks - MLP using nnet



In [148]:

    
model <- mlp_nnet_model <-  class_mlp_nnet(iris_train_n$data, "Species")
head(model$train$metrics)
test <- mlp_nnet_test <- class_test(model, iris_test_n$data, "Species")
head(test$metrics)
head(test$conf_mat)









    



# weights:  27
initial  value 101.348687 
iter  10 value 28.894876
iter  20 value 11.790441
iter  30 value 10.362204
iter  40 value 9.923474
iter  50 value 9.768647
iter  60 value 9.680004
iter  70 value 9.622098
iter  80 value 9.619029
iter  90 value 9.618843
iter 100 value 9.618835
final  value 9.618835 
converged






    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9888889 0.9833333 0.9833333 0.9916667 0.9833333 0.9833333









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9555556 0.9333333 0.9333333 0.9666667 0.9333333 0.9333333









    





       predictions
targets  1  2  3
      1  8  0  0
      2  0 11  2
      3  0  0  9

Neural Networks - MLP using RSNNS



In [149]:

    
model <- mlp_rsnns_model <- class_mlp_RSNNS(iris_train_n$data, "Species")
head(model$train$metrics)
test <- class_test(model, iris_test_n$data, "Species")
head(test$metrics)
head(test$conf_mat)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9944444 0.9916667 0.9916667 0.9958333 0.9916667 0.9916667









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9555556 0.9333333 0.9333333 0.9666667 0.9333333 0.9333333









    





       predictions
targets  1  2  3
      1  8  0  0
      2  0 11  2
      3  0  0  9

Neural Networks - RBF using RSNNS



In [150]:

    
model <- rbf_rsnns_model <- class_rbf_RSNNS(iris_train_n$data, "Species")
head(model$train$metrics)
test <- class_test(model, iris_test_n$data, "Species")
head(test$metrics)
head(test$conf_mat)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9833333 0.975 0.975 0.9875 0.975 0.975









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9111111 0.8666667 0.8666667 0.9333333 0.8666667 0.8666667









    





       predictions
targets  1  2  3
      1  8  0  0
      2  1 10  2
      3  0  1  8

Creating a SVM with RBF kernel



In [151]:

    
model <- svm_rbf_model <- class_svm_rbf(iris_train_n$data, "Species")
head(model$train$metrics)
test <- class_test(model, iris_test_n$data, "Species")
head(test$metrics)
head(test$conf_mat)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9833333 0.975 0.975 0.9875 0.975 0.975









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9555556 0.9333333 0.9333333 0.9666667 0.9333333 0.9333333









    





       predictions
targets  1  2  3
      1  8  0  0
      2  0 11  2
      3  0  0  9

Creating a SVM with polynomial kernel



In [152]:

    
model <- svm_poly_model <- class_svm_poly(iris_train_n$data, "Species")
head(model$train$metrics)
test <- class_test(model, iris_test_n$data, "Species")
head(test$metrics)
head(test$conf_mat)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9888889 0.9833333 0.9833333 0.9916667 0.9833333 0.9833333









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9555556 0.9333333 0.9333333 0.9666667 0.9333333 0.9333333









    





       predictions
targets  1  2  3
      1  8  0  0
      2  0 12  1
      3  0  1  8

Creating a SVM with sigmoid kernel



In [153]:

    
model <- svm_sigmoid_model <-class_svm_sigmoid(iris_train_n$data, "Species")
head(model$train$metrics)
test <- class_test(model, iris_test_n$data, "Species")
head(test$metrics)
head(test$conf_mat)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.85 0.775 0.775 0.8875 0.775 0.775









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.8 0.7 0.7 0.85 0.7 0.7









    





       predictions
targets 1 2 3
      1 7 1 0
      2 1 8 4
      3 0 3 6

knn prediction



In [154]:

    
model <- knn_model <-class_knn(iris_train_n$data, "Species", k=3)
head(model$train$metrics)
test <- class_test(model, iris_test_n$data, "Species")
head(test$metrics)
head(test$conf_mat)









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9833333 0.975 0.975 0.9875 0.975 0.975









    





A data.frame: 1 × 6

	accuracy f1 sensitivity specificity precision recall
	<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>


	0.9111111 0.8666667 0.8666667 0.9333333 0.8666667 0.8666667









    





       predictions
targets  1  2  3
      1  8  0  0
      2  0 11  2
      3  0  2  7

Measuring the quality of predictions



In [156]:

    
plot_size(4, 3)
zr_rocr <- compute_rocr(ZeroRule_test$predictions, ZeroRule_test$values)
plot(zr_rocr)

mlp_nnet_rocr <- compute_rocr(mlp_nnet_test$predictions, ZeroRule_test$values)
plot(mlp_nnet_rocr)



In [ ]:



In [ ]:

Sepal.Length	Sepal.Width	Petal.Length	Petal.Width	Species
<dbl>	<dbl>	<dbl>	<dbl>	<fct>
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