Users can manage their data via the “Dataset Storage” tab.
Click the “Dataset Storage” tab on the main screen to access the folder list. Neuton will show the dataset storage screen:
Dataset Storage
By default, “Dataset Storage” is empty and populates with new folders during the work in Neuton. A user can expand any folder to view the data located inside by clicking on the folder icon:
Dataset Storage. Expanded View.
A user can expand any folder by clicking the right arrow near the folder icon. In Dataset Storage, a user can see the list of solution folders, numbered in consecutive order with a Solution_ID (Solution_ID is a unique solution identifier used in Neuton).
A user can find the Solution_ID under the solution name on the “My Solutions” page.
The [Solution_ID] folder has the following structure:
artifact
stores downloadable version of a solution
dict
stores dictionaries created by Neuton for model preparation and feature engineering. These dictionaries are used during model training and prediction with the trained model.
eda
stores all information for EDA visualization
fim
stores Feature Importance Matrix for solution
predictions
stores solution web predictions you have made. Subfolders are named after the corresponding solution name. Subfolders contain CSV files with the predictions. File names inside of subfolders are formed via the following mask:[predict_<epochtime>]. You can view prediction results using the “lens” icon
processed
stores the processed dataset used for the model training
temp
stores files with service information about the model. These files store model metrics, neural network parameters and metadata
Within the dataset management tab, you can monitor your storage size usage. For example, “Usage 3.23 GB” means that 3.23 GB of Neuton storage is used for your data, models, prediction results, etc.
Dataset Management
Datasets that were uploaded to the platform for training, validation, or prediction are placed in the storage list under the folders that store training artifacts.
Datasets can be managed with the following actions:
- delete dataset
- view dataset
- download dataset
For example, to permanently delete a dataset, you can use the “Delete” icon right next to the dataset name.
Delete File
To download a dataset, use the download icon right next to the “lens” icon which is used for dataset review.
