我運行下麵的python代碼從一個磚的培訓材料。
進口mlflow mlflow進口。從pyspark.ml火花。回歸從pyspark.ml進口LinearRegression。從pyspark進口VectorAssembler特性。毫升從pyspark.ml進口管道。評估導入RegressionEvaluator mlflow.start_run (run_name =“LR-Single-Feature”)運行:# Define管道vec_assembler = VectorAssembler (inputCols =“臥室”,outputCol =“特性”)lr = LinearRegression (featuresCol =“特性”,labelCol =“價格”)管道=管道(階段= [vec_assembler, lr]) pipeline_model = pipeline.fit (train_df) # mlflow日誌參數。log_param mlflow(“標簽”,“價格”)。log_param(“特性”、“臥室”)# mlflow.spark日誌模式。log_model (pipeline_model,“模式”,input_example = train_df.limit (5) .toPandas ())
代碼的最後一行“mlflow.spark。log_model (pipeline_model,“模式”,input_example = train_df.limit (5) .toPandas ())“造成以下警告。
mlflow.utils警告。環境:遇到一個意想不到的錯誤而推斷pip需求(模式URI: / tmp / tmpchgj6je8味道:火花),跌回返回(“pyspark = = 3.3.0”)。日誌級別設置為調試看到完整的回溯。
誰能幫忙的原因和方法解決嗎?非常感謝!
