{"id":26274,"date":"2024-01-25T12:16:34","date_gmt":"2024-01-25T04:16:34","guid":{"rendered":"http:\/\/www.biocloudservice.com\/wordpress\/?p=26274"},"modified":"2024-01-25T12:16:36","modified_gmt":"2024-01-25T04:16:36","slug":"testing-based-procedures%e5%87%8f%e5%b0%91%e5%8f%98%e9%87%8f%e7%ae%80%e5%8c%96%e6%a8%a1%e5%9e%8b%e9%a2%84%e6%b5%8b%e7%bb%93%e6%9e%9c%e7%9a%84%e5%a5%bd%e5%b8%ae%e6%89%8b","status":"publish","type":"post","link":"http:\/\/www.biocloudservice.com\/wordpress\/?p=26274","title":{"rendered":"testing-based procedures\u51cf\u5c11\u53d8\u91cf\u7b80\u5316\u6a21\u578b\u9884\u6d4b\u7ed3\u679c\u7684\u597d\u5e2e\u624b"},"content":{"rendered":"

\u5728\u7edf\u8ba1\u5047\u8bbe\u68c0\u9a8c\u4e2d\uff0ctesting-based procedures\u662f\u4e00\u7c7b\u5e38\u7528\u7684\u65b9\u6cd5\uff0c\u7528\u4e8e\u5728\u751f\u7269\u4fe1\u606f\u6570\u636e\u5b66\u4e2d\u8fdb\u884c\u7279\u5f81\u9009\u62e9\u548c\u53d8\u91cf\u9009\u62e9\u3002\u8fd9\u4e9b\u65b9\u6cd5\u65e8\u5728\u4ece\u5927\u91cf\u7684\u751f\u7269\u4fe1\u606f\u6570\u636e\u4e2d\u8bc6\u522b\u51fa\u6700\u91cd\u8981\u3001\u6700\u76f8\u5173\u7684\u7279\u5f81\u6216\u53d8\u91cf\uff0c\u4ee5\u6784\u5efa\u6700\u4f73\u7684\u9884\u6d4b\u6a21\u578b\u6216\u89e3\u91ca\u751f\u7269\u5b66\u73b0\u8c61\u3002<\/p>\n

\u5728\u751f\u7269\u4fe1\u606f\u5b66\u4e2d\uff0c\u5e38\u5e38\u9762\u5bf9\u5927\u91cf\u7684\u57fa\u56e0\u8868\u8fbe\u6570\u636e\u3001\u86cb\u767d\u8d28\u5e8f\u5217\u6570\u636e\u6216\u5176\u4ed6\u751f\u7269\u5b66\u7279\u5f81\u6570\u636e\u3002\u8fd9\u4e9b\u6570\u636e\u901a\u5e38\u5305\u542b\u5927\u91cf\u7684\u53d8\u91cf\uff0c\u4f46\u5e76\u4e0d\u662f\u6240\u6709\u7684\u53d8\u91cf\u90fd\u5bf9\u6240\u7814\u7a76\u7684\u751f\u7269\u5b66\u95ee\u9898\u5177\u6709\u91cd\u8981\u5f71\u54cd\u3002\u56e0\u6b64\uff0c\u901a\u8fc7testing-based procedures\uff0c\u53ef\u4ee5\u5e2e\u52a9\u7b5b\u9009\u51fa\u90a3\u4e9b\u4e0e\u6240\u5173\u6ce8\u751f\u7269\u5b66\u7279\u6027\u5bc6\u5207\u76f8\u5173\u7684\u53d8\u91cf\uff0c\u4ece\u800c\u51cf\u5c11\u6a21\u578b\u590d\u6742\u6027\uff0c\u63d0\u9ad8\u9884\u6d4b\u548c\u89e3\u91ca\u7684\u51c6\u786e\u6027\u3002<\/p>\n

testing-based procedures\u4e2d\u7684\u4e00\u4e9b\u5e38\u89c1\u65b9\u6cd5\u5305\u62ecbackward selection\uff08\u9010\u6b65\u540e\u5411\u9009\u62e9\uff09\u3001forward selection\uff08\u9010\u6b65\u524d\u5411\u9009\u62e9\uff09\u548cstepwise regression\uff08\u9010\u6b65\u56de\u5f52\uff09\u3002\u9010\u6b65\u540e\u5411\u9009\u62e9\u4ece\u5305\u542b\u6240\u6709\u9884\u6d4b\u53d8\u91cf\u7684\u6a21\u578b\u5f00\u59cb\uff0c\u901a\u8fc7\u9010\u6b65\u5254\u9664\u90a3\u4e9b\u5bf9\u7814\u7a76\u7ed3\u679c\u5f71\u54cd\u8f83\u5c0f\u7684\u53d8\u91cf\uff0c\u4ece\u800c\u9010\u6e10\u6784\u5efa\u4e00\u4e2a\u7b80\u5316\u7684\u6a21\u578b\u3002\u9010\u6b65\u524d\u5411\u9009\u62e9\u5219\u4ece\u4e00\u4e2a\u7a7a\u6a21\u578b\u5f00\u59cb\uff0c\u9010\u6b65\u6dfb\u52a0\u5bf9\u76ee\u6807\u53d8\u91cf\u6709\u663e\u8457\u5f71\u54cd\u7684\u53d8\u91cf\u3002\u800c\u9010\u6b65\u56de\u5f52\u662f\u5c06\u9010\u6b65\u524d\u5411\u9009\u62e9\u548c\u9010\u6b65\u540e\u5411\u9009\u62e9\u7ed3\u5408\u5728\u4e00\u8d77\uff0c\u6bcf\u6b65\u8003\u8651\u540c\u65f6\u6dfb\u52a0\u548c\u5220\u9664\u53d8\u91cf\u3002<\/p>\n

\u8fd9\u4e9b\u65b9\u6cd5\u901a\u5e38\u4f1a\u57fa\u4e8e\u7edf\u8ba1\u5b66\u6307\u6807\uff08\u5982p\u503c\u3001AIC\u3001BIC\u7b49\uff09\u6765\u8bc4\u4f30\u6bcf\u4e2a\u53d8\u91cf\u7684\u91cd\u8981\u6027\u548c\u663e\u8457\u6027\u3002\u901a\u8fc7\u53cd\u590d\u6dfb\u52a0\u6216\u5220\u9664\u53d8\u91cf\uff0c\u6700\u7ec8\u5f97\u5230\u4e00\u4e2a\u8f83\u4e3a\u7b80\u5316\u7684\u6a21\u578b\uff0c\u5176\u4e2d\u5305\u542b\u6700\u91cd\u8981\u7684\u9884\u6d4b\u53d8\u91cf\uff0c\u80fd\u591f\u66f4\u597d\u5730\u89e3\u91ca\u548c\u9884\u6d4b\u751f\u7269\u5b66\u73b0\u8c61\u3002<\/p>\n

1. \u9010\u6b65\u540e\u5411\u9009\u62e9\uff08Backward Selection\uff09\uff1a<\/p>\n

\u9010\u6b65\u540e\u5411\u9009\u62e9\u662f\u4e00\u79cd\u7279\u5f81\u9009\u62e9\u7684\u65b9\u6cd5\uff0c\u5b83\u4ece\u5305\u542b\u6240\u6709\u9884\u6d4b\u53d8\u91cf\u7684\u5b8c\u6574\u6a21\u578b\u5f00\u59cb\uff0c\u5e76\u901a\u8fc7\u9010\u6b65\u5254\u9664\u5bf9\u76ee\u6807\u53d8\u91cf\u5f71\u54cd\u8f83\u5c0f\u7684\u53d8\u91cf\u6765\u6784\u5efa\u4e00\u4e2a\u7b80\u5316\u7684\u6a21\u578b\u3002\u5176\u6b65\u9aa4\u5982\u4e0b\uff1a<\/p>\n

a. \u9996\u5148\uff0c\u5c06\u6240\u6709\u9884\u6d4b\u53d8\u91cf\u5305\u542b\u5728\u6a21\u578b\u4e2d\u3002<\/p>\n

b. \u7136\u540e\uff0c\u901a\u8fc7\u4e00\u5b9a\u7684\u51c6\u5219\uff08\u5982p\u503c\u3001AIC\u3001BIC\u7b49\uff09\u6765\u68c0\u9a8c\u6bcf\u4e2a\u53d8\u91cf\u7684\u663e\u8457\u6027\uff0c\u5254\u9664\u5176\u4e2d\u5bf9\u76ee\u6807\u53d8\u91cf\u5f71\u54cd\u6700\u4e0d\u663e\u8457\u7684\u4e00\u4e2a\u3002<\/p>\n

c. \u91cd\u590d\u6b65\u9aa4b\uff0c\u76f4\u5230\u5269\u4e0b\u7684\u53d8\u91cf\u90fd\u88ab\u8ba4\u4e3a\u662f\u663e\u8457\u7684\u6216\u6ee1\u8db3\u67d0\u4e2a\u9884\u8bbe\u7684\u505c\u6b62\u51c6\u5219\u3002<\/p>\n

\u793a\u4f8b\uff1a<\/p>\n

\u6211\u4eec\u5c06\u4f7f\u7528R\u4e2d\u7684”iris”\u6570\u636e\u96c6\uff0c\u8fd9\u662f\u4e00\u4e2a\u5305\u542b\u4e86150\u4e2a\u9e22\u5c3e\u82b1\u6837\u672c\u7684\u6570\u636e\u96c6\uff0c\u5176\u4e2d\u6bcf\u4e2a\u6837\u672c\u6709\u56db\u4e2a\u7279\u5f81\uff1a\u82b1\u843c\u957f\u5ea6\uff08Sepal.Length\uff09\u3001\u82b1\u843c\u5bbd\u5ea6\uff08Sepal.Width\uff09\u3001\u82b1\u74e3\u957f\u5ea6\uff08Petal.Length\uff09\u548c\u82b1\u74e3\u5bbd\u5ea6\uff08Petal.Width\uff09\u3002\u6211\u4eec\u5c06\u5c1d\u8bd5\u4f7f\u7528\u9010\u6b65\u540e\u5411\u9009\u62e9\u6765\u5efa\u7acb\u4e00\u4e2a\u56de\u5f52\u6a21\u578b\uff0c\u5176\u4e2d\u4ee5\u82b1\u74e3\u5bbd\u5ea6\uff08Petal.Width\uff09\u4f5c\u4e3a\u76ee\u6807\u53d8\u91cf\uff0c\u5176\u4ed6\u7279\u5f81\u4f5c\u4e3a\u9884\u6d4b\u53d8\u91cf\u3002<\/p>\n

# \u52a0\u8f7d\u5fc5\u8981\u7684\u5e93<\/p>\n

> library(stats)<\/p>\n

# \u52a0\u8f7d\u6570\u636e\u96c6<\/p>\n

> data(iris)<\/p>\n

# \u521b\u5efa\u7ebf\u6027\u56de\u5f52\u6a21\u578b\uff08\u5b8c\u6574\u6a21\u578b\uff09\uff0c\u5c06Petal.Width\u4f5c\u4e3a\u76ee\u6807\u53d8\u91cf\uff0c\u5176\u4ed6\u6570\u503c\u578b\u53d8\u91cf\u4f5c\u4e3a\u9884\u6d4b\u53d8\u91cf<\/p>\n

> full_model <- lm(Petal.Width ~ ., data = iris)<\/p>\n

# \u6267\u884c\u9010\u6b65\u540e\u5411\u9009\u62e9<\/p>\n

> backward_selection <- step(full_model, direction = “backward”)<\/p>\n

Start: AIC=-531.74<\/p>\n

Petal.Width ~ Sepal.Length + Sepal.Width + Petal.Length + Species<\/p>\n

Df Sum of Sq RSS AIC<\/p>\n

<none> 3.9976 -531.74<\/p>\n

– Sepal.Length 1 0.12062 4.1182 -529.28<\/p>\n

– Petal.Length 1 0.68266 4.6802 -510.09<\/p>\n

– Sepal.Width 1 0.71403 4.7116 -509.09<\/p>\n

– Species 2 1.38273 5.3803 -491.18<\/p>\n

# \u8f93\u51fa\u9010\u6b65\u540e\u5411\u9009\u62e9\u7ed3\u679c<\/p>\n

> summary(backward_selection)<\/p>\n

Call:<\/p>\n

lm(formula = Petal.Width ~ Sepal.Length + Sepal.Width + Petal.Length +<\/p>\n

Species, data = iris)<\/p>\n

Residuals:<\/p>\n

Min 1Q Median 3Q Max<\/p>\n

-0.59239 -0.08288 -0.01349 0.08773 0.45239<\/p>\n

Coefficients:<\/p>\n

Estimate Std. Error t value Pr(>|t|)<\/p>\n

(Intercept) -0.47314 0.17659 -2.679 0.00824 **<\/p>\n

Sepal.Length -0.09293 0.04458 -2.084 0.03889 *<\/p>\n

Sepal.Width 0.24220 0.04776 5.072 1.20e-06 ***<\/p>\n

Petal.Length 0.24220 0.04884 4.959 1.97e-06 ***<\/p>\n

Speciesversicolor 0.64811 0.12314 5.263 5.04e-07 ***<\/p>\n

Speciesvirginica 1.04637 0.16548 6.323 3.03e-09 ***<\/p>\n

—<\/p>\n

Signif. codes: 0 \u2018***\u2019 0.001 \u2018**\u2019 0.01 \u2018*\u2019 0.05 \u2018.\u2019 0.1 \u2018 \u2019 1<\/p>\n

Residual standard error: 0.1666 on 144 degrees of freedom<\/p>\n

Multiple R-squared: 0.9538, Adjusted R-squared: 0.9522<\/p>\n

F-statistic: 594.9 on 5 and 144 DF, p-value: < 2.2e-16<\/p>\n

# \u7ed8\u5236\u56de\u5f52\u8bca\u65ad\u56fe<\/p>\n

> par(mfrow=c(2,2)) # \u8bbe\u7f6e\u591a\u56fe\u7ed8\u5236\u53c2\u6570<\/p>\n

> plot(backward_selection) # \u7ed8\u5236\u56de\u5f52\u8bca\u65ad\u56fe<\/p>\n

\"52326d29-1357-47a3-a28c-63b07569b28f\"<\/p>\n

\u5728\u4e0a\u8ff0\u4ee3\u7801\u4e2d\uff0c\u6211\u4eec\u52a0\u8f7d\u4e86stats\u5e93\u5e76\u52a0\u8f7d\u4e86”iris”\u6570\u636e\u96c6\u3002\u7136\u540e\uff0c\u6211\u4eec\u521b\u5efa\u4e86\u4e00\u4e2a\u5305\u542b\u6240\u6709\u6570\u503c\u578b\u53d8\u91cf\u7684\u7ebf\u6027\u56de\u5f52\u6a21\u578b\uff08\u5b8c\u6574\u6a21\u578b\uff09\u3002\u63a5\u7740\uff0c\u901a\u8fc7step\u51fd\u6570\u6267\u884c\u9010\u6b65\u540e\u5411\u9009\u62e9\uff0c\u5e76\u5c06\u9009\u62e9\u7ed3\u679c\u4fdd\u5b58\u5728backward_selection\u4e2d\u3002\u8981\u53ef\u89c6\u5316\u9010\u6b65\u540e\u5411\u9009\u62e9\u7684\u7ed3\u679c\uff0c\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528plot\u51fd\u6570\u7ed8\u5236\u6a21\u578b\u7684\u8bca\u65ad\u56fe\u3002<\/p>\n

\u8fd9\u4e2a\u793a\u4f8b\u6f14\u793a\u4e86\u5982\u4f55\u4f7f\u7528\u9010\u6b65\u540e\u5411\u9009\u62e9\u65b9\u6cd5\u6765\u5efa\u7acb\u4e00\u4e2a\u56de\u5f52\u6a21\u578b\uff0c\u4f7f\u7528\u57fa\u56e0\u8868\u8fbe\u6570\u636e\u96c6\u7684\u7279\u5f81\u4f5c\u4e3a\u9884\u6d4b\u53d8\u91cf\uff0c\u5176\u4e2d\u82b1\u74e3\u5bbd\u5ea6\uff08Petal.Width\uff09\u662f\u76ee\u6807\u53d8\u91cf\u3002\u60a8\u53ef\u4ee5\u6839\u636e\u60a8\u7684\u5177\u4f53\u751f\u7269\u5b66\u6570\u636e\u66ff\u6362\u76ee\u6807\u53d8\u91cf\u548c\u9884\u6d4b\u53d8\u91cf\uff0c\u4ee5\u5728\u5b9e\u9645\u5e94\u7528\u4e2d\u8fdb\u884c\u7279\u5f81\u9009\u62e9\u548c\u56de\u5f52\u5efa\u6a21\u3002<\/p>\n

2. \u9010\u6b65\u524d\u5411\u9009\u62e9\uff08Forward Selection\uff09\uff1a<\/p>\n

\u9010\u6b65\u524d\u5411\u9009\u62e9\u662f\u53e6\u4e00\u79cd\u7279\u5f81\u9009\u62e9\u7684\u65b9\u6cd5\uff0c\u5b83\u4e0e\u9010\u6b65\u540e\u5411\u9009\u62e9\u76f8\u53cd\uff0c\u4ece\u5305\u542b\u96f6\u4e2a\u9884\u6d4b\u53d8\u91cf\u7684\u7a7a\u6a21\u578b\u5f00\u59cb\uff0c\u9010\u6b65\u6dfb\u52a0\u5bf9\u76ee\u6807\u53d8\u91cf\u5f71\u54cd\u663e\u8457\u7684\u53d8\u91cf\u3002\u5176\u6b65\u9aa4\u5982\u4e0b\uff1a<\/p>\n

a. \u9996\u5148\uff0c\u5c06\u4e00\u4e2a\u7a7a\u6a21\u578b\u4f5c\u4e3a\u8d77\u59cb\u70b9\u3002<\/p>\n

b. \u7136\u540e\uff0c\u9010\u4e2a\u5c06\u6bcf\u4e2a\u9884\u6d4b\u53d8\u91cf\u6dfb\u52a0\u5230\u6a21\u578b\u4e2d\uff0c\u5e76\u4f7f\u7528\u4e00\u5b9a\u7684\u51c6\u5219\uff08\u5982p\u503c\u3001AIC\u3001BIC\u7b49\uff09\u6765\u68c0\u9a8c\u8be5\u53d8\u91cf\u7684\u663e\u8457\u6027\u3002<\/p>\n

c. \u91cd\u590d\u6b65\u9aa4b\uff0c\u6bcf\u6b21\u6dfb\u52a0\u5bf9\u76ee\u6807\u53d8\u91cf\u5f71\u54cd\u6700\u663e\u8457\u7684\u4e00\u4e2a\u53d8\u91cf\uff0c\u76f4\u5230\u8fdb\u4e00\u6b65\u6dfb\u52a0\u53d8\u91cf\u4e0d\u518d\u663e\u8457\u6216\u6ee1\u8db3\u67d0\u4e2a\u9884\u8bbe\u7684\u505c\u6b62\u51c6\u5219\u3002<\/p>\n

\u793a\u4f8b\uff1a<\/p>\n

\u6211\u4eec\u9700\u8981\u52a0\u8f7d\u6570\u636e\u96c6\u5e76\u6267\u884c\u9010\u6b65\u524d\u5411\u9009\u62e9\u3002\u5728\u8fd9\u4e2a\u793a\u4f8b\u4e2d\uff0c\u6211\u4eec\u5c06\u4f7f\u7528”diagnosis”\u4f5c\u4e3a\u76ee\u6807\u53d8\u91cf\uff08\u54cd\u5e94\u53d8\u91cf\uff09\uff0c\u5176\u4ed6\u6570\u503c\u578b\u53d8\u91cf\u4f5c\u4e3a\u9884\u6d4b\u53d8\u91cf\u3002<\/p>\n

# \u52a0\u8f7d\u5fc5\u8981\u7684\u5e93<\/p>\n

> library(stats)<\/p>\n

> library(caret)<\/p>\n

# \u52a0\u8f7d\u6570\u636e\u96c6<\/p>\n

> data(breast_cancer)<\/p>\n

# \u521b\u5efa\u76ee\u6807\u53d8\u91cf\u548c\u9884\u6d4b\u53d8\u91cf\u7684\u6570\u636e\u6846<\/p>\n

> data_df <- data.frame(diagnosis = breast_cancer$diagnosis,<\/p>\n

breast_cancer[, 3:12]) # \u9009\u62e93\u523012\u5217\u7684\u6570\u636e\u4f5c\u4e3a\u9884\u6d4b\u53d8\u91cf<\/p>\n

# \u5c06\u76ee\u6807\u53d8\u91cf\u8f6c\u6362\u4e3a\u4e8c\u8fdb\u5236\u56e0\u5b50\uff08\u56e0\u4e3a\u6211\u4eec\u8981\u8fdb\u884c\u4e8c\u5143\u5206\u7c7b\uff09<\/p>\n

> data_df$diagnosis <- factor(data_df$diagnosis, levels = c(“B”, “M”),<\/p>\n

labels = c(“Benign”, “Malignant”))<\/p>\n

# \u4f7f\u7528\u9010\u6b65\u524d\u5411\u9009\u62e9\u6784\u5efa\u7ebf\u6027\u56de\u5f52\u6a21\u578b<\/p>\n

> forward_selection <- train(diagnosis ~ ., data = data_df,<\/p>\n

method = “lmStepAIC”, direction = “forward”)<\/p>\n

# \u8f93\u51fa\u9010\u6b65\u524d\u5411\u9009\u62e9\u7ed3\u679c<\/p>\n

> summary(forward_selection$finalModel)<\/p>\n

# \u7ed8\u5236\u62df\u5408\u56fe<\/p>\n

> plot(forward_selection$finalModel)<\/p>\n

\"Screenshot<\/p>\n

# \u7ed8\u5236\u663e\u8457\u6027\u56fe<\/p>\n

> par(mfrow=c(1,1))<\/p>\n

> stepAIC::plot(forward_selection$finalModel)<\/p>\n

\"ols_step_forward_aic-1\"<\/p>\n

\u5728\u4e0a\u8ff0\u4ee3\u7801\u4e2d\uff0c\u6211\u4eec\u9996\u5148\u52a0\u8f7d\u4e86stats\u548ccaret\u5e93\uff0c\u5e76\u52a0\u8f7d\u4e86”breast_cancer”\u6570\u636e\u96c6\u3002\u7136\u540e\uff0c\u6211\u4eec\u521b\u5efa\u4e86\u4e00\u4e2a\u5305\u542b\u76ee\u6807\u53d8\u91cf\u548c\u9884\u6d4b\u53d8\u91cf\u7684\u6570\u636e\u6846”data_df”\u3002\u63a5\u7740\uff0c\u6211\u4eec\u5c06\u76ee\u6807\u53d8\u91cf”diagnosis”\u8f6c\u6362\u4e3a\u4e8c\u8fdb\u5236\u56e0\u5b50\uff0c\u4ee5\u4fbf\u8fdb\u884c\u4e8c\u5143\u5206\u7c7b\u3002\u4f7f\u7528train\u51fd\u6570\uff0c\u6211\u4eec\u6267\u884c\u4e86\u9010\u6b65\u524d\u5411\u9009\u62e9\uff0c\u5e76\u5c06\u9009\u62e9\u7ed3\u679c\u4fdd\u5b58\u5728forward_selection\u4e2d\u3002\u5728\u8fd9\u91cc\uff0c\u6211\u4eec\u4f7f\u7528lmStepAIC\u4f5c\u4e3a\u65b9\u6cd5\u6765\u8fdb\u884c\u9010\u6b65\u524d\u5411\u9009\u62e9\uff0c\u5e76\u6307\u5b9a\u65b9\u5411\u4e3a”forward”\uff0c\u8868\u793a\u9010\u6b65\u6dfb\u52a0\u663e\u8457\u7684\u53d8\u91cf\u3002\u8981\u53ef\u89c6\u5316\u9010\u6b65\u524d\u5411\u9009\u62e9\u7684\u7ed3\u679c\uff0c\u6211\u4eec\u53ef\u4ee5\u7ed8\u5236\u6a21\u578b\u7684\u62df\u5408\u60c5\u51b5\uff0c\u4ee5\u53ca\u53d8\u91cf\u7684\u663e\u8457\u6027\u3002<\/p>\n

3. \u9010\u6b65\u56de\u5f52\uff08Stepwise Regression\uff09\uff1a<\/p>\n

\u9010\u6b65\u56de\u5f52\u662f\u5c06\u9010\u6b65\u524d\u5411\u9009\u62e9\u548c\u9010\u6b65\u540e\u5411\u9009\u62e9\u7ed3\u5408\u5728\u4e00\u8d77\u7684\u65b9\u6cd5\u3002\u5b83\u901a\u8fc7\u5728\u6bcf\u4e2a\u6b65\u9aa4\u4e2d\u540c\u65f6\u8003\u8651\u6dfb\u52a0\u548c\u5220\u9664\u53d8\u91cf\uff0c\u6765\u6784\u5efa\u6700\u4f18\u7684\u6a21\u578b\u3002\u5176\u6b65\u9aa4\u5982\u4e0b\uff1a<\/p>\n

a. \u9996\u5148\uff0c\u5c06\u4e00\u4e2a\u7a7a\u6a21\u578b\u4f5c\u4e3a\u8d77\u59cb\u70b9\u3002<\/p>\n

b. \u7136\u540e\uff0c\u6309\u7167\u4e00\u5b9a\u7684\u51c6\u5219\uff08\u5982p\u503c\u3001AIC\u3001BIC\u7b49\uff09\u8003\u8651\u6dfb\u52a0\u548c\u5220\u9664\u53d8\u91cf\u3002<\/p>\n

c. \u5728\u6bcf\u4e2a\u6b65\u9aa4\u4e2d\uff0c\u53ef\u4ee5\u540c\u65f6\u6dfb\u52a0\u4e00\u4e2a\u53d8\u91cf\u548c\u5220\u9664\u4e00\u4e2a\u4e0d\u663e\u8457\u7684\u53d8\u91cf\u3002<\/p>\n

d. \u91cd\u590d\u6b65\u9aa4b\u548cc\uff0c\u76f4\u5230\u8fdb\u4e00\u6b65\u6dfb\u52a0\u548c\u5220\u9664\u53d8\u91cf\u90fd\u4e0d\u518d\u663e\u8457\u6216\u6ee1\u8db3\u67d0\u4e2a\u9884\u8bbe\u7684\u505c\u6b62\u51c6\u5219\u3002<\/p>\n

\u793a\u4f8b\uff1a<\/p>\n

\u6211\u4eec\u9700\u8981\u52a0\u8f7d\u201cmtcars\u201d\u6570\u636e\u96c6\u5e76\u6267\u884c\u9010\u6b65\u56de\u5f52\uff0c\u5728\u8fd9\u4e2a\u793a\u4f8b\u4e2d\uff0c\u6211\u4eec\u5c06\u4f7f\u7528\u201cmpg\u201d\u4f5c\u4e3a\u76ee\u6807\u53d8\u91cf\uff08\u54cd\u5e94\u53d8\u91cf\uff09\uff0c\u5176\u4ed6\u6570\u503c\u578b\u53d8\u91cf\u4f5c\u4e3a\u9884\u6d4b\u53d8\u91cf\u3002<\/p>\n

# \u52a0\u8f7d\u5fc5\u8981\u7684\u5e93<\/p>\n

> library(stats)<\/p>\n

# \u52a0\u8f7d\u6570\u636e\u96c6<\/p>\n

> data(mtcars)<\/p>\n

# \u521b\u5efa\u76ee\u6807\u53d8\u91cf\u548c\u9884\u6d4b\u53d8\u91cf\u7684\u6570\u636e\u6846<\/p>\n

> data_df <- data.frame(mpg = mtcars$mpg,<\/p>\n

+ mtcars[, c(2:4, 6:7)]) # \u9009\u62e9\u7b2c2\u81f34\u5217\u548c\u7b2c6\u81f37\u5217\u7684\u6570\u636e\u4f5c\u4e3a\u9884\u6d4b\u53d8\u91cf<\/p>\n

# \u4f7f\u7528\u9010\u6b65\u56de\u5f52\u6784\u5efa\u7ebf\u6027\u56de\u5f52\u6a21\u578b<\/p>\n

> stepwise_model <- step(lm(mpg ~ ., data = data_df), direction = “both”)<\/p>\n

Start: AIC=65.19<\/p>\n

mpg ~ cyl + disp + hp + wt + qsec<\/p>\n

Df Sum of Sq RSS AIC<\/p>\n

– qsec 1 1.759 170.44 63.526<\/p>\n

– disp 1 6.534 175.22 64.410<\/p>\n

– hp 1 6.983 175.67 64.492<\/p>\n

<none> 168.69 65.194<\/p>\n

– cyl 1 16.950 185.63 66.258<\/p>\n

– wt 1 73.848 242.53 74.813<\/p>\n

Step: AIC=63.53<\/p>\n

mpg ~ cyl + disp + hp + wt<\/p>\n

Df Sum of Sq RSS AIC<\/p>\n

– disp 1 6.176 176.62 62.665<\/p>\n

<none> 170.44 63.526<\/p>\n

– hp 1 18.048 188.49 64.746<\/p>\n

+ qsec 1 1.759 168.69 65.194<\/p>\n

– cyl 1 24.546 194.99 65.831<\/p>\n

– wt 1 90.925 261.37 75.206<\/p>\n

Step: AIC=62.66<\/p>\n

mpg ~ cyl + hp + wt<\/p>\n

Df Sum of Sq RSS AIC<\/p>\n

<none> 176.62 62.665<\/p>\n

– hp 1 14.551 191.17 63.198<\/p>\n

+ disp 1 6.176 170.44 63.526<\/p>\n

– cyl 1 18.427 195.05 63.840<\/p>\n

+ qsec 1 1.401 175.22 64.410<\/p>\n

– wt 1 115.354 291.98 76.750<\/p>\n

# \u8f93\u51fa\u9010\u6b65\u56de\u5f52\u7ed3\u679c<\/p>\n

> summary(stepwise_model)<\/p>\n

Call:<\/p>\n

lm(formula = mpg ~ cyl + hp + wt, data = data_df)<\/p>\n

Residuals:<\/p>\n

Min 1Q Median 3Q Max<\/p>\n

-3.9290 -1.5598 -0.5311 1.1850 5.8986<\/p>\n

Coefficients:<\/p>\n

Estimate Std. Error t value Pr(>|t|)<\/p>\n

(Intercept) 38.75179 1.78686 21.687 < 2e-16 ***<\/p>\n

cyl -0.94162 0.55092 -1.709 0.098480 .<\/p>\n

hp -0.01804 0.01188 -1.519 0.140015<\/p>\n

wt -3.16697 0.74058 -4.276 0.000199 ***<\/p>\n

—<\/p>\n

Signif. codes: 0 \u2018***\u2019 0.001 \u2018**\u2019 0.01 \u2018*\u2019 0.05 \u2018.\u2019 0.1 \u2018 \u2019 1<\/p>\n

Residual standard error: 2.512 on 28 degrees of freedom<\/p>\n

Multiple R-squared: 0.8431, Adjusted R-squared: 0.8263<\/p>\n

F-statistic: 50.17 on 3 and 28 DF, p-value: 2.184e-11<\/p>\n

# \u7ed8\u5236\u56de\u5f52\u8bca\u65ad\u56fe<\/p>\n

> par(mfrow=c(2,2)) # \u8bbe\u7f6e\u591a\u56fe\u7ed8\u5236\u53c2\u6570<\/p>\n

> plot(stepwise_model) # \u7ed8\u5236\u56de\u5f52\u8bca\u65ad\u56fe<\/p>\n

\"d39e1880-c57d-4d63-bdc0-435304c7551e\"<\/p>\n

\u5728\u4e0a\u8ff0\u4ee3\u7801\u4e2d\uff0c\u6211\u4eec\u52a0\u8f7d\u4e86stats\u5e93\uff0c\u5e76\u52a0\u8f7d\u4e86”mtcars”\u6570\u636e\u96c6\u3002\u7136\u540e\uff0c\u6211\u4eec\u521b\u5efa\u4e86\u4e00\u4e2a\u5305\u542b\u76ee\u6807\u53d8\u91cf\u548c\u9884\u6d4b\u53d8\u91cf\u7684\u6570\u636e\u6846”data_df”\u3002\u63a5\u7740\uff0c\u6211\u4eec\u4f7f\u7528lm\u51fd\u6570\u521b\u5efa\u4e86\u4e00\u4e2a\u7ebf\u6027\u56de\u5f52\u6a21\u578b\uff08\u5b8c\u6574\u6a21\u578b\uff09\u3002\u7136\u540e\uff0c\u901a\u8fc7step\u51fd\u6570\u6267\u884c\u9010\u6b65\u56de\u5f52\uff0c\u5e76\u5c06\u9009\u62e9\u7ed3\u679c\u4fdd\u5b58\u5728stepwise_model\u4e2d\u3002\u8981\u53ef\u89c6\u5316\u9010\u6b65\u56de\u5f52\u7684\u7ed3\u679c\uff0c\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528plot\u51fd\u6570\u7ed8\u5236\u6a21\u578b\u7684\u8bca\u65ad\u56fe\u3002<\/p>\n

\u8fd9\u4e2a\u793a\u4f8b\u6f14\u793a\u4e86\u5982\u4f55\u4f7f\u7528\u9010\u6b65\u56de\u5f52\u65b9\u6cd5\u6765\u5efa\u7acb\u4e00\u4e2a\u56de\u5f52\u6a21\u578b\uff0c\u4f7f\u7528\u4e0e\u751f\u7269\u5b66\u76f8\u5173\u7684\u6570\u636e\u96c6\u7279\u5f81\u4f5c\u4e3a\u9884\u6d4b\u53d8\u91cf\uff0c\u5176\u4e2d”mpg”\u662f\u76ee\u6807\u53d8\u91cf\u3002<\/p>\n

\u4ee5\u4e0a\u5c31\u662f\u5bf9\u5047\u8bbe\u6d4b\u8bd5\u68c0\u9a8c\u5bf9\u8fc7\u7a0b\u7684\u7b80\u5355\u4ecb\u7ecd\u5566\uff0c\u9700\u8981\u6ce8\u610f\u7684\u662f\uff0c\u867d\u7136testing-based procedures\u662f\u4e00\u79cd\u5e38\u7528\u7684\u7279\u5f81\u9009\u62e9\u65b9\u6cd5\uff0c\u4f46\u5728\u5e94\u7528\u65f6\u9700\u8981\u8c28\u614e\u3002\u8fc7\u5ea6\u4f7f\u7528\u8fd9\u4e9b\u65b9\u6cd5\u53ef\u80fd\u4f1a\u5bfc\u81f4\u8fc7\u62df\u5408\u95ee\u9898\uff0c\u5373\u6a21\u578b\u5728\u8bad\u7ec3\u6570\u636e\u4e0a\u8868\u73b0\u826f\u597d\uff0c\u4f46\u5728\u672a\u77e5\u6570\u636e\u4e0a\u8868\u73b0\u8f83\u5dee\u3002\u56e0\u6b64\uff0c\u7814\u7a76\u4eba\u5458\u9700\u8981\u5728\u4f7f\u7528\u8fd9\u4e9b\u65b9\u6cd5\u65f6\u8fdb\u884c\u9002\u5f53\u7684\u9a8c\u8bc1\u548c\u4ea4\u53c9\u9a8c\u8bc1\uff0c\u4ee5\u786e\u4fdd\u6240\u9009\u53d6\u7684\u53d8\u91cf\u5728\u4e0d\u540c\u6570\u636e\u96c6\u4e0a\u90fd\u80fd\u591f\u7a33\u5b9a\u5730\u53d1\u6325\u4f5c\u7528\uff0c\u5e76\u5f97\u5230\u53ef\u9760\u7684\u751f\u7269\u5b66\u89e3\u91ca\u3002<\/p>\n

\u5c0f\u4f19\u4f34\u4eec\uff0c\u4eca\u5929\u6709\u6ca1\u6709\u5b66\u5230\u65b0\u77e5\u8bc6\u5462\uff0c\u60f3\u8981\u7ee7\u7eed\u4e86\u89e3R\u8bed\u8a00\u5185\u5bb9\u53ef\u4ee5\u6301\u7eed\u5173\u6ce8\u5c0f\u679c\u54e6\uff5e\uff5e\u6216\u8005\u4e5f\u53ef\u4ee5\u5173\u6ce8\u6211\u4eec\u7684\u5b98\u7f51\u4e5f\u4f1a\u6301\u7eed\u66f4\u65b0\u7684\u54e6\uff5e http:\/\/www.biocloudservice.com\/home.html<\/a><\/p>\n

References:<\/p>\n

    \n
  1. https:\/\/rpubs.com\/ricardosc\/breast-cancer<\/a><\/li>\n
  2. https:\/\/www.google.com\/url?sa=i&url=https%3A%2F%2Folsrr.rsquaredacademy.com%2Freference%2Fols_step_forward_aic.html&psig=AOvVaw2tgi8QZIbdfegRyXLinm-9&ust=1690626806590000&source=images&cd=vfe&opi=89978449&ved=0CBAQjRxqFwoTCPiWvOeZsYADFQAAAAAdAAAAABAJ<\/a><\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"

    \u5728\u7edf\u8ba1\u5047\u8bbe\u68c0\u9a8c\u4e2d\uff0ctesting-based procedures\u662f\u4e00\u7c7b\u5e38\u7528\u7684\u65b9\u6cd5\uff0c\u7528\u4e8e\u5728\u751f\u7269\u4fe1\u606f\u6570\u636e\u5b66\u4e2d\u8fdb\u884c […]<\/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\/26274"}],"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=26274"}],"version-history":[{"count":1,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/26274\/revisions"}],"predecessor-version":[{"id":26279,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/26274\/revisions\/26279"}],"wp:attachment":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=26274"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=26274"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=26274"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}