2021-02-02 12:00:41 +00:00
|
|
|
|
---
|
|
|
|
|
toc_priority: 222
|
|
|
|
|
---
|
|
|
|
|
|
|
|
|
|
# stochasticLogisticRegression {#agg_functions-stochasticlogisticregression}
|
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
该函数实现随机逻辑回归。 它可以用于二进制分类问题,支持与stochasticLinearRegression相同的自定义参数,并以相同的方式工作。
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
### 参数 {#agg_functions-stochasticlogisticregression-parameters}
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
参数与stochasticLinearRegression中的参数完全相同:
|
2021-02-02 12:00:41 +00:00
|
|
|
|
`learning rate`, `l2 regularization coefficient`, `mini-batch size`, `method for updating weights`.
|
2021-03-09 11:20:52 +00:00
|
|
|
|
欲了解更多信息,参见 [参数] (#agg_functions-stochasticlinearregression-parameters).
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
``` sql
|
2021-02-02 12:00:41 +00:00
|
|
|
|
stochasticLogisticRegression(1.0, 1.0, 10, 'SGD')
|
|
|
|
|
```
|
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
**1.** 拟合
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
|
|
|
|
<!-- -->
|
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
参考[stochasticLinearRegression](#stochasticlinearregression-usage-fitting) `拟合` 章节文档。
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
预测标签的取值范围为\[-1, 1\]
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
**2.** 预测
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
|
|
|
|
<!-- -->
|
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
使用已经保存的state我们可以预测标签为 `1` 的对象的概率。
|
2021-02-02 12:00:41 +00:00
|
|
|
|
``` sql
|
|
|
|
|
WITH (SELECT state FROM your_model) AS model SELECT
|
|
|
|
|
evalMLMethod(model, param1, param2) FROM test_data
|
|
|
|
|
```
|
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
查询结果返回一个列的概率。注意 `evalMLMethod` 的第一个参数是 `AggregateFunctionState` 对象,接下来的参数是列的特性。
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
我们也可以设置概率的范围, 这样需要给元素指定不同的标签。
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
|
|
|
|
``` sql
|
|
|
|
|
SELECT ans < 1.1 AND ans > 0.5 FROM
|
|
|
|
|
(WITH (SELECT state FROM your_model) AS model SELECT
|
|
|
|
|
evalMLMethod(model, param1, param2) AS ans FROM test_data)
|
|
|
|
|
```
|
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
结果是标签。
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
`test_data` 是一个像 `train_data` 一样的表,但是不包含目标值。
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
**参见**
|
2021-02-02 12:00:41 +00:00
|
|
|
|
|
2021-03-09 11:20:52 +00:00
|
|
|
|
- [随机指标线性回归](../../../sql-reference/aggregate-functions/reference/stochasticlinearregression.md#agg_functions-stochasticlinearregression)
|
|
|
|
|
- [线性回归和逻辑回归之间的差异](https://stackoverflow.com/questions/12146914/what-is-the-difference-between-linear-regression-and-logistic-regression)
|