sfaira.train.SummarizeGridsearchCelltype.plot_best_classwise_heatmap¶
- SummarizeGridsearchCelltype.plot_best_classwise_heatmap(organ: str, organism: str, datapath: str, store_format: str, targetpath: str, configpath: str, partition_select: str = 'val', metric_select: str = 'custom_cce_agg', metric_show: str = 'f1', collapse_cv: str = 'mean', min_cells: int = 10, height_fig: int = 7, width_fig: int = 7)¶
Plot evaluation metric heatmap for specified organ by cell classes and model types.
- Parameters
organ – Organ to plot in heatmap.
organism – Species that the gridsearch was run on
datapath – Path to the local sfaira data repository
store_format –
targetpath –
configpath –
partition_select – Based on which partition to select the best model - train - val - test - all
metric_select – Based on which metric to select the best model - loss - accuracy - custom_cce_agg - acc_agg - f1 - tpr - fpr
metric_show – Which classwise metric to plot. - accuracy - f1
collapse_cv – How to collapse over the single cv runs.
min_cells – Minimum number of cells of a type must be present in the whole dataset for that class to be included in the plot.
height_fig – Figure height.
width_fig – Figure width.
- Returns
fig, axs, sns_data_heatmap