https://kedro.org/ logo
Join Slack
Powered by
Hi everyone, I'm using Kedro 0.17.4 and I'm having...
# questions
r
Hi everyone, I'm using Kedro 0.17.4 and I'm having this issue:
ValueError: The following datasets are used with transcoding, but were referenced without the separator: delisting.item_features. Please specify a transcoding option or rename the datasets
It comes from the __validate_transcoded_inputs_outputs_
d
What are all the catalog entries/references you have for 
item_features
?
r
Copy code
_pandas_parquet: &_pandas_parquet
  type: pandas.ParquetDataSet

_spark_parquet: &_spark_parquet
  type: spark.SparkDataSet
  file_format: parquet
  load_args:
    header: true
  save_args:
    header: true
    mode: overwrite

delisting.basic_item_features:
  <<: *_spark_parquet
  filepath: ${base_path_spark}/feature_generation/basic_item_features.parquet

delisting.basket_level_item_features:
  <<: *_spark_parquet
  filepath: ${base_path_spark}/feature_generation/basket_level_item_features.parquet

delisting.customer_level_item_features:
  <<: *_spark_parquet
  filepath: ${base_path_spark}/feature_generation/customer_level_item_features.parquet

delisting.item_features@spark:
  <<: *_spark_parquet
  filepath: ${base_path_spark}/feature_generation/item_features.parquet

delisting.item_features@pandas:
  <<: *_pandas_parquet
  filepath: ${base_path}/feature_generation/item_features.parquet
This is my catalog, but IDK if the error is referring to the path or the name of the dataset
d
When you create a node in the pipeline, have you also used the transcoded names? It seems like that's what may be missing
i.e. make sure you have 
node(my_node_func, "delisting.item_features@pandas", "whatever")
instead of 
node(my_node_func, "delisting.item_features", "whatever")
👍 1
r
Cool! Yes I was using the wrong dataset name on other pipeline definition
👍 1
13 Views
Open in Slack
PreviousNext
Powered by