Ch-2

library(keras)

# Data
data <- read.csv('https://raw.githubusercontent.com/bkrai/DeepLearningR/master/Cardiotocographic.csv')
data$NSP <- data$NSP -1
barplot(prop.table(table(data$NSP)),
        col = rainbow(3),
        ylim = c(0, 0.8),
        ylab = 'Proportion',
        xlab = 'NSP',
        cex.names = 1.5)

# Matrix
data <- as.matrix(data)
dimnames(data) <- NULL  

# Normalize 
data[,1:21] <- normalize(data[,1:21])
data[,22] <- as.numeric(data[,22]) -1
summary(data)

# Partition
set.seed(1234)
ind <- sample(2, nrow(data), replace = T, prob=c(.7, .3))
training <- data[ind==1, 1:21]
test <- data[ind==2, 1:21]
trainingtarget <- data[ind==1, 22]
testtarget <- data[ind==2, 22]

# One hot encoding 
trainLabels <- to_categorical(trainingtarget)
testLabels <- to_categorical(testtarget)

# Creat sequential model and add layers 
model <- keras_model_sequential()
model %>%  layer_dense(units = 8, activation = 'relu', input_shape = c(21)) %>%   
  layer_dense(units = 3, activation = 'softmax') 

# Compile 
model %>% compile(loss = 'categorical_crossentropy', 
  optimizer = 'adam',
  metrics = 'accuracy')

# Fit 
model_one <- model %>%   
  fit(training, 
  trainLabels, 
  epochs = 200,
  batch_size = 32, 
  validation_split = 0.2)

# Evaluate 
model %>%  evaluate(test, testLabels)  

# Prediction 
pred <- model %>% 
  predict_classes(test)
table(Predicted=pred, Actual=testtarget)

# Fine-tune 
model <- keras_model_sequential()
model %>%   layer_dense(units = 8, activation = 'relu', input_shape = c(21)) %>%  
  layer_dense(units = 5, activation = 'relu') %>% 
  layer_dense(units = 3, activation = 'softmax') 

model %>% 
  compile(loss = 'categorical_crossentropy', 
  optimizer = 'adam',
  metrics = 'accuracy')

history <- model %>%   
  fit(training, 
      trainLabels, 
      epochs = 200,
      batch_size = 32, 
      validation_split = 0.2)

pred <- model %>% 
  predict_classes(test)
table(Predicted=pred, Actual=testtarget)

# Fine-tune - add units
model <- keras_model_sequential()
model %>% layer_dense(units = 40, activation = 'relu', input_shape = c(21)) %>%  
         layer_dropout(rate = 0.4) %>%
         layer_dense(units = 30, activation = 'relu') %>%
         layer_dropout(rate = 0.3) %>%
         layer_dense(units = 20, activation = 'relu') %>%
         layer_dropout(rate = 0.2) %>%
         layer_dense(units = 3, activation = 'softmax')  
model %>% 
  compile(loss = 'categorical_crossentropy', 
          optimizer = 'adam',
          metrics = 'accuracy')
history <- model %>%   
  fit(training, 
      trainLabels, 
      epochs = 100,
      batch_size = 32,  
      validation_split = 0.2,
      class_weight = list("0"=1,"1"=5.6, "2" = 9.4))

pred <- model %>% 
  predict_classes(test)
table(Predicted=pred, Actual=testtarget)