線性學習程式回應格式

JSON 回應格式

所有 Amazon SageMaker 內建演算法都遵循一般資料格式-推論中所述的通用輸入推論格式。以下是 SageMaker 線性學習器演算法的可用輸出格式。

二元分類

let response =   {
    "predictions":    [
        {
            "score": 0.4,
            "predicted_label": 0
        } 
    ]
}

多類別分類

let response =   {
    "predictions":    [
        {
            "score": [0.1, 0.2, 0.4, 0.3],
            "predicted_label": 2
        } 
    ]
}

迴歸

let response =   {
    "predictions":    [
        {
            "score": 0.4
        } 
    ]
}

JSONLINES 回應格式

二元分類

{"score": 0.4, "predicted_label": 0}

多類別分類

{"score": [0.1, 0.2, 0.4, 0.3], "predicted_label": 2}

迴歸

{"score": 0.4}

RECORDIO 回應格式

二元分類

[
    Record = {
        features = {},
        label = {
            'score': {
                keys: [],
                values: [0.4]  # float32
            },
            'predicted_label': {
                keys: [],
                values: [0.0]  # float32
            }
        }
    }
]

多類別分類

[
    Record = {
    "features": [],
    "label":    {
            "score":  {
                    "values":   [0.1, 0.2, 0.3, 0.4]   
            },
            "predicted_label":  {
                    "values":   [3]
            }
       },
    "uid":  "abc123",
    "metadata": "{created_at: '2017-06-03'}"
   }
]

迴歸

[
    Record = {
        features = {},
        label = {
            'score': {
                keys: [],
                values: [0.4]  # float32
            }   
        }
    }
]