Function Reference (English)¶
One file per function/method.
- analyze_abf_to_event_df
- build_best_model
- build_DL_model
- butterworth_filtfilt
- classify_new_samples
- classify_new_samples_DL compatibility note
- create_analysis_object
- current
- denoise
- detect_events
- detect_events_simple
- detect_events_cusum
- detect_events_hmm
- detect_events_pelt
- detect_events_threshold
- do_pca
- do_tsne
- do_umap
- estimate_baseline
- events_to_dataframe
- event_current
- event_current_label
- event_current_simple
- extract_features
- filter_events
- load
- load_model_package
- median_filter
- model_cm
- model_metric_bar
- plot_fold_loss
- moving_average
- stacked_bar
- box_significance
- plot_2d
- plot_3d
- predict_events
- preprocess_signal
- preview_signal
- read_abf
- read_abf_all
- read_csv
- remove_slow_drift
- rolling_quantile
- save_model_package
- select_feature_columns
- train_event_classifier
- visualize_signal
Current DL names¶
MultiSampleAnalysis.build_DL_model supports the built-in residual-CNN, MAGJAM, and MAGJAM_d4 names. MAGJAM is the d8 waveform-only model and MAGJAM_d4 is the d4 waveform-only model. interp_length defaults to 500; interp_method defaults to "interp" and also accepts "padding".
There is one unified prediction method, classify_new_samples, for both ML and DL. The similarly named file above is retained only to point older users to that method; no separate classify_new_samples_DL method exists.