All Machine Learning Challenge tasks
Machine Learning Challenge
This is a standard machine learning challenge with a hidden private dataset.
It offers a labeled training set and an unlabeled test set.
The task is to label the unlabeled instances. Only the OpenML server knows the correct labels, and will evaluate the submitted predictions using these hidden labels. The evaluation procedure, measure, and cost function (if any) are provided.
Inputs
| cost_matrix |
CostMatrix |
A matrix describing the cost of miss-classifications per type. |
optional |
| estimation_procedure |
Estimation Procedure |
The estimation procedure used to validate the generated models |
required |
| evaluation_measures |
String |
The evaluation measures to optimize for, e.g., cpu time, accuracy |
optional |
| source_data |
Dataset |
The input data for this task |
required |
| source_data_labeled |
Dataset |
The labelled version of the dataset |
required |
| target_feature |
String |
The name of the dataset feature to be used as the target feature. |
required |
Outputs
| evaluations |
KeyValue |
A list of user-defined evaluations of the task as key-value pairs. |
optional |
| model |
File |
A file containing the model built on all the input data. |
optional |
| predictions |
Predictions |
The desired output format |
optional |
Attribution
| Author(s) | "Jan van Rijn","Joaquin Vanschoren" |
| Contributor(s) | |