使用R中的keras使用Pima印度糖尿病数据集拟合Conv1d对糖尿病患者进行分类的误差

library(keras)

# 导入Pima印度糖尿病数据集
data <- read.csv("pima_diabetes.csv")

# 数据预处理
# 将特征值和标签分开
features <- data[, 1:8]
labels <- data[, 9]

# 数据归一化
features <- scale(features)

# 划分训练集和测试集
set.seed(123)
train_indices <- sample(1:nrow(data), 0.8 * nrow(data))
train_features <- features[train_indices, ]
train_labels <- labels[train_indices]
test_features <- features[-train_indices, ]
test_labels <- labels[-train_indices]

# 创建Conv1d模型
model <- keras_model_sequential()
model %>%
  layer_conv_1d(filters = 32, kernel_size = 3, activation = "relu", input_shape = c(8, 1)) %>%
  layer_flatten() %>%
  layer_dense(units = 1, activation = "sigmoid")

# 编译模型
model %>% compile(
  loss = "binary_crossentropy",
  optimizer = optimizer_rmsprop(),
  metrics = c("accuracy")
)

# 将数据reshape为Conv1D所需的形状
train_features <- array_reshape(train_features, c(dim(train_features), 1))
test_features <- array_reshape(test_features, c(dim(test_features), 1))

# 训练模型
model %>% fit(
  x = train_features, y = train_labels,
  epochs = 10, batch_size = 32,
  validation_data = list(test_features, test_labels)
)

# 评估模型
metrics <- model %>% evaluate(
  x = test_features, y = test_labels
)

cat("准确率:", metrics$accuracy, "\n")
cat("精确率:", metrics$precision, "\n")
cat("召回率:", metrics$recall, "\n")
cat("F1分数:", metrics$f1_score, "\n")