Convert raw dataset to D3M dataset

Currently d3m package needs Python 3.6 only.

pip install d3m
python <train_data.csv> <test_data.csv> <label> <metric> -t classification <-t ...>


Some examples of valid commands are -

python train_data.csv test_data.csv Label accuracy -t classification
python train_data.csv test_data.csv Value meanSquaredError -t regression

-t option should be used to specify task types(s), data types(s). metrics. This script will create a directory structure “raw” for your dataset in D3M format. This dataset should be used as input to ./scripts/

This is the structure created for a generated D3M dataset:

raw$ tree
├── TEST
│   ├── dataset_TEST
│   │   ├── datasetDoc.json
│   │   ├── metadata.json
│   │   └── tables
│   │       └── learningData.csv
│   └── problem_TEST
│       └── problemDoc.json
    ├── dataset_TRAIN
    │   ├── datasetDoc.json
    │   ├── metadata.json
    │   └── tables
    │       └── learningData.csv
    └── problem_TRAIN
        └── problemDoc.json

8 directories, 8 files

Example of creating D3M dataset for image regression

python train.csv test.csv WRISTBREADTH meanSquaredError -t regression -t image
Namespace(dataFileName='train.csv', metric='meanSquaredError', target='WRISTBREADTH', tasks=['regression', 'image'], testDataFileName='test.csv')
Going to create TRAIN files!
Going to create TEST files!
Please enter directory name for TRAIN media files: train_images
Please enter directory name for TEST media files: test_images
Please enter column name for media files: image_file

Note: Some task/data type(s) may not be entirely automated (Eg., object detection, graph problems). TRAIN, TEST hierarchies will be made available. However, datasetDoc.json might need to be customized for linking resources/tables for the specific task. For this purpose, example datasets are provided for reference purposes.

Valid task types(s)

linkPrediction, graphMatching, forecasting, classification, semiSupervised, clustering, collaborativeFiltering, regression, objectDetection, vertexNomination, communityDetection, vertexClassification

Valid data type(s)

Valid data type(s) to specify are- audio, image, video, text, timeSeries

Valid metrics

classification/linkPrediction/graphMatching/vertexNomination/vertexClassification: accuracy, f1Macro, f1Micro, rocAuc, rocAucMacro, rocAucMicro regression/forecasting/collaborativeFiltering: rSquared, meanSquaredError, meanSquaredError, meanAbsoluteError communityDetection/clustering: normalizedMutualInformation

Sample D3M dataset(s) for task type(s), data types(s):