Repository navigation
Load data into LearningPipeline without TextLoader #336
Description
Question:
So I'm actually capturing my data directly from an array of sensors, and these are currently in memory. I don't have millions of datapoints, just a few hundred. I do not want to write them to file. Is TextLoader is the only way as of now to ingest data to LearningPipeline?
I wondering if I could point to an array of objects instead of a text file to load them up, is there any alternate way I could ingest data?
Activity
rlingineni commentedon Jun 8, 2018
I think I might have found an answer:
I might be able to use the
CollectionDataSource, seems like it's a part of the new release.Fix progress bar in linux/osx (dotnet#336)
code migration from machinelearning-automl to machinelearning repo fe…
Fix progress bar in linux/osx (dotnet#336)
Added onnx export functionality for LpNormNormalizingTransformer (#4161)
Added Onnx export functionality to PCATransformer (#4188)