WOLFRAM

Wolfram 语言与系统参考资料中心

ClassPriors

是 Classify 及相关函数的一个选项，指定输出类别假定的显式先验概率，独立于从训练集推导出的结果.

更多信息
Details and Options Details and Options
范例  
基本范例  
应用  
参见
历史
按以下格式引用：

ClassPriors

ClassPriors

是 Classify 及相关函数的一个选项，指定输出类别假定的显式先验概率，独立于从训练集推导出的结果.

更多信息

  • ClassPriors<|class1p1,class2p2,…|> 指定将特定输入归类于类 classi 中的先验概率为 pi.
  • 默认情况下，类别先验从训练集的类别频率计算.
  • 在 ClassPriors<|class1p1,class2p2,…|> 中，未指定的类别先验将被设定为它们重新缩放的默认值，以确保分布的标准化.

范例

打开所有单元 关闭所有单元

基本范例  (1)

训练不平衡训练集上的分类器：

Wolfram Language code: trainingset = {1 -> "a", 2 -> "b", 3 -> "a", 4 -> "b", 5 -> "b", 6 -> "b", 7 -> "b"};
Wolfram Language code: c = Classify[trainingset]

获得新范例的类别概率：

Wolfram Language code: c[2.5, "Probabilities"]

设置类别的均匀先验分布：

Wolfram Language code: c[2.5, "Probabilities", ClassPriors -> <|"a" -> 0.5, "b" -> 0.5|>]

设置分类器内部类别的先验分布：

Wolfram Language code: c = Classify[trainingset, ClassPriors -> <|"a" -> 0.5, "b" -> 0.5|>]
Wolfram Language code: c[2.5, "Probabilities"]

未指定类的先验是剩余概率质量的函数：

Wolfram Language code: c[2.5, "Probabilities", ClassPriors -> <|"a" -> 0.8|>]
Wolfram Language code: c[2.5, "Probabilities", ClassPriors -> <|"a" -> 1|>]

应用  (3)

将内置语言分类器限制为仅两种语言：

Wolfram Language code: Classify["Language", {"What is this language?", "¿Qué idioma es ese?", "Quelle est cette langue?"}, ClassPriors -> <|Entity["Language", "English"] -> 0.5, Entity["Language", "Spanish"] -> 0.5|>]

训练分类器来识别有机体在平衡训练组上是否健康：

Wolfram Language code: c = Classify[{{1., 3.2, "Y"} -> "Diseased", {2.4, -3.1, "N"} -> "Healthy", {-2.1, 4.2, "N"} -> "Diseased", {2.7, -10, "N"} -> "Healthy", {-4., 6.2, "Y"} -> "Diseased", {2.5, 7., "Y"} -> "Healthy"}]

求新有机体患病的概率：

Wolfram Language code: c[{0.8, 2.2, "Y"}, "Probabilities"]

求同一有机体患病的概率，结合患病有机体仅代表总体的1%这一先验知识：

Wolfram Language code: c[{0.8, 2.2, "Y"}, "Probabilities", ClassPriors -> <|"Diseased" -> 0.01|>]

训练一个分类器，使其能够根据网站用户的行为预测他是否会购买已给定物品：

Wolfram Language code: c = Classify[{ {1.7, False, 1.3, True} -> "will buy", {-4, False, 1.1, True} -> "won't buy", {5.7, True, 1.2, True} -> "will buy", {7.3, False, 1.3, False} -> "won't buy", {-2.3, False, 1.8, True} -> "won't buy", {3.2, True, 1.4, True} -> "won't buy" }]

定义并可视化模型，购买该物品的概率是外界温度的一个函数：

Wolfram Language code: prior[temp_] := 0.15 + ArcTan[-0.1 * temp] / 10;
Wolfram Language code: Plot[prior[temp], {temp, -20, 40}, PlotRange -> {0, 1}]

使用该模型和 WeatherData 提升对巴黎用户的预测效果，而无需额外的训练：

Wolfram Language code: c[{7.2, False, 1.5, True}, "Probabilities", ClassPriors -> <|"will buy" -> prior[QuantityMagnitude[WeatherData["Paris", "Temperature"]]]|>]
Wolfram Research (2014)，ClassPriors，Wolfram 语言函数，https://reference.wolfram.com/language/ref/ClassPriors.html.

文本

Wolfram Research (2014)，ClassPriors，Wolfram 语言函数，https://reference.wolfram.com/language/ref/ClassPriors.html.

CMS

Wolfram 语言. 2014. "ClassPriors." Wolfram 语言与系统参考资料中心. Wolfram Research. https://reference.wolfram.com/language/ref/ClassPriors.html.

APA

Wolfram 语言. (2014). ClassPriors. Wolfram 语言与系统参考资料中心. 追溯自 https://reference.wolfram.com/language/ref/ClassPriors.html 年

BibTeX

@misc{reference.wolfram_2026_classpriors, author="Wolfram Research", title="{ClassPriors}", year="2014", howpublished="\url{https://reference.wolfram.com/language/ref/ClassPriors.html}", note=[Accessed: 05-September-2026]}

BibLaTeX

@online{reference.wolfram_2026_classpriors, organization={Wolfram Research}, title={ClassPriors}, year={2014}, url={https://reference.wolfram.com/language/ref/ClassPriors.html}, note=[Accessed: 05-September-2026]}