Skip to main content

13  Categorical Feature Extraction

Open Data Extraction → Categorical in a local installation. Using the CSV tables saved by Numerical, FLIM Playground:

Categorical Feature Extraction with a local dataset folder, underscore delimiter, category selections, slot mappings, and a preview of the assigned labels.

Combine numerical datasets and map FOV-name segments to categorical features.

13.1 Merging Datasets

Enter the full path to the folder containing your saved numerical feature CSV files. The folder is scanned recursively for CSV files. Each usable dataset must contain the configured cell and FOV identifiers. Cell identifiers must be nonmissing and unique within the file and across the accepted files; a file containing overlapping identifiers is skipped. If no file is usable, an error says why, such as no CSV files or none with both identifiers; warnings above it explain other skipped files.

Important

The FOV identifier uniquely identifies a field of view and is assumed to contain all information necessary to extract the categorical features (e.g., treatment, time point, etc.), by including segments separated by a certain delimiter (e.g., _). Categorical features are assigned from individual segments or their combinations. Example: The FOV Panc1_Cyanide_dish1 denotes that it comes from the Panc1 cell line, under the Cyanide treatment, in the 1st dish.

Categorical feature assignment is available only when all FOV names have the same number of segments.

Before merging, all-empty columns are removed. The remaining datasets are stacked using only their common columns, and the page reports columns discarded during merging.

13.2 Assigning Categorical Features

Available categories are the configured categorical names that are not already columns in the merged dataset.

Set Field of View Name Delimiter to the separator used in the FOV names. Leaving it empty treats each complete name as one slot. For each chosen category, select one or more slots; their values are joined in the selected order.

The live preview shows one row from each of the first five FOVs. Check the assigned values before exporting.

FOV-name slots map to categorical columns, with a preview above Confirm category mapping & export the combined dataset.

Map FOV-name slots to categories and check the preview before exporting.

Click Confirm category mapping & export the combined dataset to save the merged table and assigned categories as {timestamp}_combined.csv in the input folder. Upload this CSV directly to Data Analysis in either the desktop or hosted app. The column editor opens with guessed roles and groups; check them and click Use for this upload.