Tabular Datasets

Introduction

Tabular datasets in Fed-BioMed handle structured data in tabular formats for classification and regression tasks.

Key Features: - Automatic data loading with delimiter detection - Format conversion (pandas, NumPy, PyTorch tensors) - Framework compatibility (PyTorch and scikit-learn)

Data Structure

Basic CSV: 

feature1,feature2,feature3,target
1.2,3.4,5.6,0
2.1,4.3,6.5,1

Data Preparation

  1. Clean data (handle missing values, outliers)
  2. Format as CSV for maximum compatibility
  3. Validate data types in columns

Deployment

Node-Side

Register using the Fed-BioMed node CLI, see Deploying Datasets for details.:

fedbiomed node dataset add
# 1. Select "csv"
# 2. Path: /path/to/your/data.csv
# 3. Tags: #tabular

Researcher-Side

Access tabular datasets through experiment configuration:

from fedbiomed.researcher.experiment import Experiment

# Select nodes with tabular classification datasets
experiment = Experiment(
    tags=['#tabular'],
    model=my_model,
    training_plan_class=MyTrainingPlan,
    training_args=training_args
)

Integration with Training Plans

PyTorch Training Plan for Tabular Data

...

Scikit-learn Training Plan for Tabular Data

...

Best Practices and Common Issues

  • Verify all numerical columns contain only numbers
  • Check for mixed data types in the same column
  • Handle string representations of numbers
  • Choose appropriate imputation strategies
  • Consider the impact of missing data
  • Validate features don't introduce data leakage
  • Check for infinite or NaN values