dynverse / dyneval

Metrics to compare two or more trajectories ⚖
2 stars 2 forks source link

Most of the metrics are not available. Some are always identical. #60

Closed noobCoding closed 1 year ago

noobCoding commented 1 year ago

image

Hi @rcannood,

I followed instructions to evaluate available methods at https://dynverse.org/developers/evaluating-ti-method/

As you can see, I tested with provided datasets and with 8 methods: models <- infer_trajectories(dataset, list(ti_comp1(), ti_mst(), ti_cellrouter(), ti_paga(), ti_paga_tree(), ti_monocle_ica(), ti_slingshot(), ti_dpt()))

But the output metrics ('rf_mse' and 'lm_mse') are always identical... And most of other metrics are not obtained (all 0s). Do you know what happened?