Skip to main content
op.LocalDataset is the fastest way to get local data into OuterProduct. Build one from a CSV or Parquet file (or an in-memory frame), then call .upload() to stage it through a presigned URL and get back a Dataset you can pass directly to training or inference. It is completely stateless. There is no credential to configure in the Console.

​
Supported formats

Build a LocalDataset from a local CSV or Parquet file (from_csv / from_parquet), or directly from an in-memory frame or array — pandas (from_pandas), polars (from_polars), or numpy (from_numpy). Whatever the source, .upload() serializes the rows to Parquet and stages them through the presigned URL.

​
Upload a file and train a model

1

Build a LocalDataset

Create a LocalDataset from your file with from_csv (or from_parquet).
import outerproduct as op

local = op.LocalDataset.from_csv("customers.csv")
2

Upload it

Call .upload() to stage the rows. OuterProduct uploads them via a presigned URL and returns a Dataset.
dataset = local.upload()
3

Train a model

Pass the Dataset to op.reasoning.fit() along with your label column and task type, then call .wait() to block until training completes.
model = op.reasoning.fit(dataset, task=op.Binclass(label_column="churn")).wait()

​
Complete example

import outerproduct as op

dataset = op.LocalDataset.from_csv("customers.csv").upload()

model = op.reasoning.fit(dataset, task=op.Binclass(label_column="churn")).wait()
Use a Parquet file instead of CSV when your dataset is large. Parquet files are compressed and columnar, so they upload faster and consume less bandwidth.

​
Other connectors

If your data lives in a cloud data store rather than on your local machine, use one of the cloud connectors instead: