ls_F1.Rd
ls_F1
is a function that returns a list of F1-Score
corresponding to the given list of imputed datasets.
resample_method
is needed because with 'bootstrap' method, we could
have repeated lines in the imputed datasets,
and with both 'jackknife' and 'bootstrap', the imputed datasets could not
cover all the lines.
ls_F1(
df_comp,
ls_df_imp,
mask,
col_cat_comp,
col_cat_imp,
resample_method = "bootstrap",
combine_method = "onehot",
dict_cat = NULL
)
The original complete dataset.
List of imputed dataset.
Mask of missingness (1 means missing value and 0 means observed value).
Indices of categorical columns in the complete dataset.
Indices of categorical columns in the imputed dataset.
Default value is 'bootstrap', could also be 'jackknife' or 'none'.
When resample_method
= 'bootstrap',
combine_method
could be 'factor' or 'onehot'.
When method
= 'onehot', ls_F1
takes the average of the one-hot
probability vector for each observation, then choose the position of maximum
probability as the predicted category. When method
= 'factor', or
each observation, ls_F1
chooses the mode value over the imputed
dataframes as the predicted category.
The dictionary of categorical columns names if "onehot" method is applied. For example, it could be list("Y7"=c("Y7_1","Y7_2"), "Y8"=c("Y8_1","Y8_2","Y8_3")).
list_F1
List of F1 corresponding to the given list of
imputed datasets.
Mean_F1
Mean value of F1.
Variance_F1
Variance of F1.