Перейти к основному содержимому

Export data and results

Export in the header of the data table downloads one Excel workbook with the table, its description and the results of all completed analyses of the experiment.

Experiment Data Table(9 rows × 7 columns)
|
#
SAMPLE NAME
Sample Info
A
CELL COORDINATE
Sample Info
B
Origin
Classification
C
plate_01.jpg
R
D
plate_01.jpg
G
E
plate_01.jpg
B
F
plate_02.jpg
intensity
G
1Tea AA1China182.00121.0064.00118.67
2Tea AA2China181.00120.0064.00117.33
3Tea AA3China183.00122.0065.00120.00
4Tea BB1India151.0098.0047.0096.33
5Tea BB2India150.0097.0046.0095.67
6Tea BB3India153.00100.0048.0097.33
7Tea CC1China171.00113.0059.00110.67
8Tea CC2China170.00111.0058.00109.00
9Tea CC3China173.00114.0059.00111.33
9 rows × 7 columns |4 data channels |3 unique samples
Export is the green button in the header of the data table.

Exporting the data table​

  1. Click Save Changes at the top of the experiment page if you edited the table. The export is built on the server from the saved table; unsaved edits are not included.
  2. Click Export. The button shows Exporting... while the file is prepared.
  3. The browser downloads <experiment name>_data.xlsx.

The workbook always has these sheets:

SheetContent
DataThe data table: one row per row of the table, columns in the same order. Headers are the internal column names: sample_name, cell_coordinate, class_<type>, <photo>_<channel> (for example plate_01.jpg_R). Values are not rounded.
Column_MetadataOne row per column: Column Name, Type (sample_info, class, data), Source, Job ID, Channel, Filename, Calculate Type, Job Order, Class Type.
Experiment_InfoExperiment Name, Experiment ID, Created, Rows, Columns, Generation Mode, Plate Rows, Plate Columns, Samples Count, Total Analyses, Completed Analyses.

The Data sheet can be edited in Excel and loaded back with Import.

Exporting analysis summaries​

When the experiment has completed analyses, the workbook has two more sheets.

Analyses_Summary — the accuracy table of the results summary, one row per combination of settings:

ColumnContent
Label Columnlabel column of the analyses
Data Rangedata range of the analyses
Included Classesclasses used, or All classes
Descriptionyour description of the combination from the summary
one column per method (LDA, PLSDA, …)cross-validated accuracy from 0 to 1

If a method was run several times with the same settings, each run gets its own row, in the order the runs were made.

Analyses_Details — one block per completed analysis, placed side by side. Each block contains, depending on the method:

  • Basic Information: Analysis ID, Analysis Name, Analysis Type, Label Column, Data Range, Created, CV Method, CV Folds, Binary Classification, and Feature Selection with Selected Features Count if features were selected;
  • Key Model Parameters;
  • Cross-Validation Metrics: mean and standard deviation of accuracy, precision, recall, F1 score, sensitivity, specificity and AUC over the folds (see metrics);
  • Fold Accuracies (accuracy of every fold);
  • Average Confusion Matrix;
  • Per-Class Metrics (Full Model): precision, recall, F1-score and support of every class for the model trained on all data;
  • Transformed Data (Scores) and Loadings for PCA (PC), LDA (LD) and PLS-DA (LV);
  • Permutation Plots: the p-value and the CV p-value of every permutation test.

Analyses that are still running or that failed are not exported.