{"id":24321,"date":"2024-01-11T17:09:23","date_gmt":"2024-01-11T09:09:23","guid":{"rendered":"http:\/\/www.biocloudservice.com\/wordpress\/?p=24321"},"modified":"2024-01-11T17:09:25","modified_gmt":"2024-01-11T09:09:25","slug":"%e6%9c%aa%e6%9d%a5%e6%97%a9%e7%9f%a5%e9%81%93%ef%bc%9a%e7%94%a8keras%e7%9a%84%e6%b7%b1%e5%ba%a6%e5%ad%a6%e4%b9%a0%e9%a2%84%e6%b5%8b%e7%96%be%e7%97%85","status":"publish","type":"post","link":"http:\/\/www.biocloudservice.com\/wordpress\/?p=24321","title":{"rendered":"\u672a\u6765\u65e9\u77e5\u9053\uff1a\u7528Keras\u7684\u6df1\u5ea6\u5b66\u4e60\u9884\u6d4b\u75be\u75c5"},"content":{"rendered":"

\u6df1\u5ea6\u5b66\u4e60\u5c31\u50cf\u662f\u673a\u5668\u5b66\u4e60\u7684\u70ab\u9177\u7248\u672c\uff01\u5b83\u662f\u4e00\u79cd\u8ba9\u673a\u5668\u53d8\u5f97\u8d85\u7ea7\u806a\u660e\u7684\u9ed1\u79d1\u6280\u3002\u901a\u8fc7\u6a21\u62df\u4eba\u8111\u795e\u7ecf\u7f51\u7edc\u7684\u65b9\u5f0f\uff0c\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u53ef\u4ee5\u4ece\u5927\u91cf\u6570\u636e\u4e2d\u5b66\u4e60\u5e76\u505a\u51fa\u51c6\u786e\u7684\u9884\u6d4b\u3002\u5c31\u50cf\u5c0f\u5b9d\u5b9d\u5b66\u4e60\u8d70\u8def\u4e00\u6837\uff0c\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u4e5f\u662f\u901a\u8fc7\u4e0d\u65ad\u7ec3\u4e60\u548c\u8bd5\u9519\u6765\u63d0\u5347\u81ea\u5df1\u7684\u80fd\u529b\u3002\u5b83\u53ef\u4ee5\u8bc6\u522b\u56fe\u50cf\u3001\u7ffb\u8bd1\u8bed\u8a00\u3001\u81ea\u52a8\u9a7e\u9a76\uff0c\u751a\u81f3\u53ef\u4ee5\u4e3a\u6211\u4eec\u5199\u8bd7\uff01\u603b\u4e4b\uff0c\u6df1\u5ea6\u5b66\u4e60\u8ba9\u673a\u5668\u53d8\u5f97\u50cf\u4eba\u4e00\u6837\u806a\u660e\uff0c\u5e26\u6765\u4e86\u65e0\u6570\u6709\u8da3\u548c\u5b9e\u7528\u7684\u5e94\u7528\u3002<\/p>\n

Keras\uff08https:\/\/keras.io\/\uff09\u5c31\u50cf\u662f\u6df1\u5ea6\u5b66\u4e60\u7684\u745e\u58eb\u519b\u5200\uff01\u5b83\u662f\u4e00\u4e2a\u9177\u70ab\u4e14\u597d\u7528\u7684\u5de5\u5177\uff0c\u53ef\u4ee5\u5e2e\u52a9\u4f60\u6784\u5efa\u548c\u8bad\u7ec3\u795e\u7ecf\u7f51\u7edc\u3002\u4e0d\u7ba1\u4f60\u662f\u4e00\u4e2a\u65b0\u624b\u8fd8\u662f\u4e00\u4e2a\u4e13\u4e1a\u4eba\u58eb\uff0cKeras\u90fd\u4f1a\u7ed9\u4f60\u5e26\u6765\u60ca\u559c\uff01\u60f3\u8c61\u4e00\u4e0b\uff0c\u4f60\u60f3\u8981\u5efa\u7acb\u4e00\u4e2a\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\uff0c\u4f46\u662f\u4e0d\u60f3\u4ece\u5934\u5f00\u59cb\u5199\u4e00\u5927\u5806\u7e41\u7410\u7684\u4ee3\u7801\u3002\u8fd9\u65f6\u5019\uff0cKeras\u5c31\u6d3e\u4e0a\u4e86\u7528\u573a\uff01\u5b83\u63d0\u4f9b\u4e86\u4e00\u79cd\u7b80\u5355\u800c\u5f3a\u5927\u7684\u65b9\u5f0f\u6765\u521b\u5efa\u795e\u7ecf\u7f51\u7edc\u3002\u4f60\u53ea\u9700\u8981\u4e00\u4e9b\u7b80\u5355\u7684\u4ee3\u7801\uff0c\u5c31\u53ef\u4ee5\u50cf\u642d\u79ef\u6728\u4e00\u6837\u7ec4\u88c5\u4f60\u7684\u6a21\u578b\u3002<\/p>\n

\u4e0d\u4ec5\u5982\u6b64\uff0cKeras\u8fd8\u6709\u4e00\u5806\u9177\u70ab\u7684\u529f\u80fd\u3002\u5b83\u652f\u6301\u5404\u79cd\u5e38\u89c1\u7684\u795e\u7ecf\u7f51\u7edc\u5c42\uff0c\u6bd4\u5982\u5168\u8fde\u63a5\u5c42\u3001\u5377\u79ef\u5c42\u3001\u5faa\u73af\u5c42\u7b49\u3002\u4f60\u53ea\u9700\u8981\u9009\u62e9\u5408\u9002\u7684\u5c42\uff0c\u7136\u540e\u5c06\u5b83\u4eec\u5806\u53e0\u5728\u4e00\u8d77\uff0c\u5c31\u80fd\u6784\u5efa\u51fa\u4f60\u60f3\u8981\u7684\u6a21\u578b\u3002\u800c\u4e14\uff0cKeras\u8fd8\u652f\u6301\u591a\u79cd\u4f18\u5316\u5668\u548c\u635f\u5931\u51fd\u6570\uff0c\u8fd9\u4e9b\u662f\u8bad\u7ec3\u6a21\u578b\u65f6\u5fc5\u4e0d\u53ef\u5c11\u7684\u4e1c\u897f\u3002\u4f60\u53ef\u4ee5\u6839\u636e\u4f60\u7684\u4efb\u52a1\u548c\u559c\u597d\u6765\u9009\u62e9\u5408\u9002\u7684\u4f18\u5316\u5668\uff0c\u8ba9\u4f60\u7684\u6a21\u578b\u8bad\u7ec3\u5f97\u66f4\u5feb\u3001\u66f4\u51c6\u786e\u3002<\/p>\n

\u6700\u9177\u7684\u662f\uff0c<\/a>Keras\u4e0d\u4ec5\u6613\u4e8e\u4f7f\u7528\uff0c\u800c\u4e14\u8fd8\u6709\u4e00\u4e2a\u5e9e\u5927\u7684\u793e\u533a\u652f\u6301\u3002\u4f60\u53ef\u4ee5\u5728\u793e\u533a\u91cc\u627e\u5230\u5404\u79cd\u5404\u6837\u7684\u4f8b\u5b50\u548c\u5e2e\u52a9\u3002\u8fd8\u6709\u5f88\u591a\u6709\u8da3\u7684\u6559\u7a0b\u548c\u535a\u5ba2\uff0c\u53ef\u4ee5\u8ba9\u4f60\u5728\u6df1\u5ea6\u5b66\u4e60\u7684\u4e16\u754c\u91cc\u7545\u6e38\u3002<\/p>\n

\u603b\u800c\u8a00\u4e4b\uff0cKeras\u5c31\u662f\u4e00\u4e2a\u9177\u70ab\u5b9e\u7528\u7684\u6df1\u5ea6\u5b66\u4e60\u5de5\u5177\uff0c\u5e2e\u52a9\u4f60\u6784\u5efa\u5f3a\u5927\u7684\u795e\u7ecf\u7f51\u7edc\u3002\u4e0d\u7ba1\u4f60\u662f\u6df1\u5ea6\u5b66\u4e60\u65b0\u624b\u8fd8\u662f\u9ad8\u624b\uff0cKeras\u90fd\u662f\u4f60\u7684\u597d\u670b\u53cb\uff0c\u8ba9\u4f60\u8f7b\u677e\u6109\u5feb\u5730\u63a2\u7d22\u6df1\u5ea6\u5b66\u4e60\u7684\u5947\u5999\u4e16\u754c\uff01<\/p>\n

Keras\u4e3b\u9875<\/strong><\/p>\n

<\/p>\n

\u4ee5\u4e0b\u662f\u4e00\u4e2a\u7b80\u5355\u7684\u793a\u4f8b\uff0c\u5c55\u793a\u5982\u4f55\u4f7f\u7528R\u5305keras\u8fdb\u884c\u7cd6\u5c3f\u75c5\u9884\u6d4b\u6a21\u578b\u7684\u8bad\u7ec3\u548c\u8bc4\u4f30\uff1a<\/p>\n

\u4ee3\u7801\u5177\u4f53\u5305\u62ec\uff1a<\/p>\n

Step1 \u5bfc\u5165\u6570\u636e\u96c6\u4ee5\u53ca\u6570\u636e\u9884\u5904\u7406<\/p>\n

# \u8fd9\u91cc\u7684”inputdata.txt”\u662f\u81ea\u884c\u51c6\u5907\u7684\u672c\u5730\u6587\u4ef6\uff0c\u5c0f\u679c\u7ed9\u5927\u5bb6\u9644\u5728\u6700\u540e\u3002<\/p>\n

# 1.\u5bfc\u5165\u6570\u636e\u96c6\u4ee5\u53ca\u6570\u636e\u9884\u5904\u7406<\/p>\n

# \u5b89\u88c5\u548c\u52a0\u8f7d\u6240\u9700\u7684\u5e93<\/p>\n

# install.packages(“keras”)<\/p>\n

library(keras)<\/p>\n

# \u5bfc\u5165\u6570\u636e\u96c6<\/p>\n

data <- read.table(file = “D:\/wanglab\/life\/ziyuan\/20230628\/inputdata.txt”,header = T) # \u66ff\u6362\u4e3a\u5b9e\u9645\u6570\u636e\u96c6\u6587\u4ef6\u7684\u8def\u5f84<\/p>\n

# \u6570\u636e\u9884\u5904\u7406<\/p>\n

# \u6570\u636e\u6807\u51c6\u5316<\/p>\n

data[, 1:8] <- scale(data[, 1:8])<\/p>\n

# \u5c06\u76ee\u6807\u53d8\u91cf\u8f6c\u6362\u4e3a\u56e0\u5b50<\/p>\n

data$diabetes <- as.factor(data$diabetes)<\/p>\n

# \u5212\u5206\u6570\u636e\u96c6\u4e3a\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6<\/p>\n

set.seed(123) # \u8bbe\u7f6e\u968f\u673a\u79cd\u5b50\uff0c\u4fdd\u8bc1\u7ed3\u679c\u53ef\u590d\u73b0<\/p>\n

indices <- sample(1:nrow(data), nrow(data)*0.8) # 80%\u7684\u6570\u636e\u7528\u4e8e\u8bad\u7ec3\uff0c20%\u7684\u6570\u636e\u7528\u4e8e\u6d4b\u8bd5<\/p>\n

train_data <- data[indices, ]<\/p>\n

test_data <- data[-indices, ]<\/p>\n

Step2\u521b\u5efa\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b<\/p>\n

# 2.<\/a>\u521b\u5efa\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b<\/p>\n

model <- keras_model_sequential() # \u521b\u5efa\u5e8f\u8d2f\u6a21\u578b<\/p>\n

model %>%<\/p>\n

layer_dense(units = 32, activation = “relu”, input_shape = 8) %>% # \u6dfb\u52a0\u4e00\u4e2a\u5168\u8fde\u63a5\u5c42\uff0c\u5305\u542b32\u4e2a\u795e\u7ecf\u5143\uff0c\u4f7f\u7528ReLU\u6fc0\u6d3b\u51fd\u6570<\/p>\n

layer_dense(units = 16, activation = “relu”) %>% # \u6dfb\u52a0\u7b2c\u4e8c\u4e2a\u5168\u8fde\u63a5\u5c42\uff0c\u5305\u542b16\u4e2a\u795e\u7ecf\u5143\uff0c\u4f7f\u7528ReLU\u6fc0\u6d3b\u51fd\u6570<\/p>\n

layer_dense(units = 1, activation = “sigmoid”) # \u6dfb\u52a0\u8f93\u51fa\u5c42\uff0c\u4f7f\u7528Sigmoid\u6fc0\u6d3b\u51fd\u6570\uff0c\u8f93\u51fa\u4e3a\u4e8c\u8fdb\u5236\u5206\u7c7b\u7ed3\u679c<\/p>\n

Step3\u7f16\u8bd1\u6a21\u578b<\/p>\n

# 3.\u7f16\u8bd1\u6a21\u578b<\/p>\n

model %>% compile(<\/p>\n

loss = “binary_crossentropy”, # \u4f7f\u7528\u4e8c\u5143\u4ea4\u53c9\u71b5\u4f5c\u4e3a\u635f\u5931\u51fd\u6570<\/p>\n

optimizer = “adam”, # \u4f7f\u7528Adam\u4f18\u5316\u5668\u8fdb\u884c\u53c2\u6570\u4f18\u5316<\/p>\n

metrics = c(“accuracy”) # \u6307\u5b9a\u8bc4\u4f30\u6307\u6807\u4e3a\u51c6\u786e\u7387<\/p>\n

)<\/p>\n

Step4\u8bad\u7ec3\u6a21\u578b<\/p>\n

# 4.\u8bad\u7ec3\u6a21\u578b<\/p>\n

history <- model %>% fit(<\/p>\n

x = as.matrix(train_data[, 1:8]), # \u8f93\u5165\u7279\u5f81<\/p>\n

y = as.numeric(train_data$diabetes) – 1, # \u76ee\u6807\u53d8\u91cf\uff08\u5c06\u56e0\u5b50\u8f6c\u6362\u4e3a\u6570\u503c\uff09<\/p>\n

epochs = 10, # \u8fed\u4ee3\u6b21\u6570<\/p>\n

batch_size = 32, # \u6279\u5904\u7406\u5927\u5c0f<\/p>\n

validation_split = 0.2 # \u5c06\u8bad\u7ec3\u96c6\u768420%\u7528\u4e8e\u9a8c\u8bc1<\/p>\n

)<\/p>\n

\u968f\u7740\u8bad\u7ec3\u6b21\u6570\u589e\u52a0\u51c6\u786e\u7387\u548c\u635f\u5931\u53d8\u5316\u6298\u7ebf\u56fe<\/strong> \"Rplot\"<\/p>\n

Step5\u8bc4\u4f30\u6a21\u578b<\/p>\n

# 5.\u8bc4\u4f30\u6a21\u578b<\/p>\n

eval <- model %>% evaluate(<\/p>\n

x = as.matrix(test_data[, 1:8]), # \u6d4b\u8bd5\u96c6\u7684\u8f93\u5165\u7279\u5f81<\/p>\n

y = as.numeric(test_data$diabetes) – 1, # \u6d4b\u8bd5\u96c6\u7684\u76ee\u6807\u53d8\u91cf\uff08\u5c06\u56e0\u5b50\u8f6c\u6362\u4e3a\u6570\u503c\uff09<\/p>\n

verbose = 0 # \u4e0d\u663e\u793a\u8be6\u7ec6\u4fe1\u606f<\/p>\n

)<\/p>\n

cat(“Test Accuracy:”, eval, “\\n”)<\/p>\n

eval[2]<\/p>\n

# \u7ed8\u5236\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u7684\u51c6\u786e\u7387\u548c\u635f\u5931\u53d8\u5316\u66f2\u7ebf<\/p>\n

plot(history)<\/p>\n

<\/a>\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u7684\u51c6\u786e\u7387\u548c\u635f\u5931\u53d8\u5316\u66f2\u7ebf<\/strong> <\/p>\n

# 6.\u7ed8\u5236AUC\u66f2\u7ebf<\/p>\n

# \u9884\u6d4b\u6982\u7387<\/p>\n

predictions <- model %>% predict(as.matrix(test_data[, 1:8]))<\/p>\n

predicted_classes <- ifelse(predictions > 0.5, 1, 0) # \u5c06\u6982\u7387\u8f6c\u6362\u4e3a\u7c7b\u522b\u9884\u6d4b\uff08\u5927\u4e8e0.5\u4e3a\u6b63\u7c7b\uff0c\u5426\u5219\u4e3a\u8d1f\u7c7b\uff09<\/p>\n

# \u8ba1\u7b97AUC<\/p>\n

library(pROC)<\/p>\n

roc_obj <- roc(test_data$diabetes, predictions)<\/p>\n

auc_value <- auc(roc_obj)<\/p>\n

# \u7ed8\u5236AUC\u66f2\u7ebf<\/p>\n

plot(roc_obj, main = “ROC Curve”)<\/p>\n

text(0.5, 0.3, paste0(“AUC = “, round(auc_value, 3)), col = “red”, cex = 1.2)<\/p>\n

AUC\u66f2\u7ebf\u56fe<\/strong><\/p>\n

<\/p>\n

\u6ce8\uff1a<\/p>\n

# \u8bf7\u786e\u4fdd\u5c06\u8def\u5f84”inputdata.txt”\u66ff\u6362\u4e3a\u5b9e\u9645\u6570\u636e\u96c6\u6587\u4ef6\u7684\u8def\u5f84\uff0c\u5e76\u6839\u636e\u9700\u8981\u8c03\u6574\u6a21\u578b\u53c2\u6570\u548c\u5176\u4ed6\u914d\u7f6e\u3002<\/p>\n

# \u5728\u7ed8\u5236AUC\u66f2\u7ebf\u4e4b\u524d\uff0c\u786e\u4fdd\u60a8\u7684\u6a21\u578b\u8f93\u51fa\u7684\u662f\u9884\u6d4b\u7684\u6982\u7387\u503c\uff0c\u800c\u4e0d\u662f\u7c7b\u522b\u6807\u7b7e\u3002<\/p>\n

\u5389\u5bb3\u5427\uff0c\u6df1\u5ea6\u5b66\u4e60\u5728R\u8bed\u8a00\u4e2d\u7684\u5b9e\u73b0\u975e\u5e38\u7b80\u5355\u3002\u6211\u4eec\u4f7f\u7528\u7684\u90fd\u662f\u5165\u95e8\u7ea7\u51fd\u6570\uff0c\u53ea\u9700\u5bf9R\u8bed\u8a00\u6709\u4e00\u5b9a\u4e86\u89e3\uff0c\u5c31\u53ef\u4ee5\u8f7b\u677e\u5b9e\u73b0\u5566\u3002<\/p>\n

\u9700\u8981\u6ce8\u610f\u7684\u662f\uff0c\u5982\u679c\u8981\u4f7f\u7528\u6df1\u5ea6\u5b66\u4e60\u8fdb\u884c\u5206\u6790\uff0c\u5e76\u4e14\u7279\u5f81\u503c\u8f83\u591a\uff0c\u53ef\u80fd\u9700\u8981\u5927\u91cf\u7684\u5185\u5b58\u3002\u5728\u672c\u5730\u7535\u8111\u4e0a\u4f1a\u9762\u4e34\u5185\u5b58\u4e0d\u8db3\u7684\u95ee\u9898\u3002\u56e0\u6b64\uff0c\u5efa\u8bae\u4f7f\u7528\u670d\u52a1\u5668\u8fdb\u884c\u5206\u6790\u3002\u7136\u800c\uff0c\u79df\u7528\u670d\u52a1\u5668\u7684\u8d39\u7528\u53ef\u80fd\u8f83\u9ad8\u3002\u4e3a\u4e86\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\uff0c\u6211\u4eec\u5efa\u8bae\u4f7f\u7528\u6211\u4eec\u7684\u4e91\u751f\u4fe1\u5e73\u53f0\uff0c\u8be5\u5e73\u53f0\u63d0\u4f9b\u5f3a\u5927\u7684\u8ba1\u7b97\u8d44\u6e90\u548c\u6df1\u5ea6\u5b66\u4e60\u5de5\u5177\u3002\u60a8\u53ef\u4ee5\u8bbf\u95ee\u6211\u4eec\u7684\u4e91\u751f\u4fe1\u5e73\u53f0\uff0c\u94fe\u63a5\u4e3a\uff1ahttp:\/\/www.biocloudservice.com\/home.html\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"

\u6df1\u5ea6\u5b66\u4e60\u5c31\u50cf\u662f\u673a\u5668\u5b66\u4e60\u7684\u70ab\u9177\u7248\u672c\uff01\u5b83\u662f\u4e00\u79cd\u8ba9\u673a\u5668\u53d8\u5f97\u8d85\u7ea7\u806a\u660e\u7684\u9ed1\u79d1\u6280\u3002\u901a\u8fc7\u6a21\u62df\u4eba\u8111\u795e\u7ecf\u7f51\u7edc\u7684\u65b9\u5f0f\uff0c\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"jetpack_post_was_ever_published":false,"_jetpack_newsletter_access":""},"categories":[1],"tags":[],"jetpack_featured_media_url":"","_links":{"self":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/24321"}],"collection":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=24321"}],"version-history":[{"count":1,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/24321\/revisions"}],"predecessor-version":[{"id":24326,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/24321\/revisions\/24326"}],"wp:attachment":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=24321"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=24321"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=24321"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}