我正在尝试使用 TARGET_TIME_SERIES 中的销售额、RELATED_TIME_SERIES 中的价格和 ITEM_METADATA 中的产品类型来训练预测 这些是我在 Featurizations 中设置的参数:
{
"AttributeName": "demand",
"FeaturizationPipeline": [
{
"FeaturizationMethodName": "filling",
"FeaturizationMethodParameters": {
"aggregation": "sum",
"backfill": "zero",
"frontfill": "none",
"middlefill": "zero"
}
}
]
},
{
"AttributeName": "promotional_price",
"FeaturizationPipeline": [
{
"FeaturizationMethodName": "filling",
"FeaturizationMethodParameters": {
"aggregation": "first",
"backfill": "median",
"frontfill": "none",
"futurefill": "max",
"middlefill": "median"
}
}
]
},
{
"AttributeName": "original_price",
"FeaturizationPipeline": [
{
"FeaturizationMethodName": "filling",
"FeaturizationMethodParameters": {
"aggregation": "first",
"backfill": "median",
"frontfill": "none",
"futurefill": "max",
"middlefill": "median"
}
}
]
}
]
但这给了我一个错误:
We were unable to train your predictor.
Invalid featurization parameters : {original_price=[frontfill=none, aggregation=first], promotional_price=[frontfill=none, aggregation=first]}.
我不明白。此参数来自文档