Communications Mining 用户指南
Precision measures the proportion of the predictions the model makes that were actually correct. This means that it identifies what proportion were true positives of all the positive predictions that the model made.
Precision = true positives / (true positives + false positives)
例如，对于数据集中每 100 条预测为具有“请求信息”标签的消息，精度指的是正确预测“信息请求”标签的次数占总预测次数的百分比。
95％ 的精度意味着每 100 条预测为具有特定标签的消息，其中 95 条会被正确标注，而 5 条会被错误地标注，这意味着它们不应该使用该标签进行标注。
For a more detailed explanation on how precision works, check Precision and recall explained.
MAP是衡量模型整体性能的最有用方法之一，也是相互比较不同模型版本的简便方法。
MAP分数取分类中每个标签的平均精度分数的平均值，该标签在“验证”使用的训练集中至少包含 20 个示例。
Typically, the higher the MAP score, the better the model is performing overall, although, this is not the only factor that should be considered when understanding how healthy a model is. It is also important to know that your model is unbiased and has high coverage.
