minor changes

fsspark/fs/multivariate.py

@@ -43,9 +43,9 @@ def multivariate_correlation_selector(fsdf: FSDataFrame,
     Compute the correlation matrix (Pearson) among input features and select those below a specified threshold.
 
     :param fsdf: Input FSDataFrame
-    :param strict: If True (default), apply hard filtering (strict) to remove highly related features.
+    :param strict: If True (default), apply hard filtering (strict) to remove highly correlated features.
                    Otherwise, find the maximal independent set of highly correlated features (approximate method).
-                   The approximate method is experimental.
+                   `Warning`: The approximate method is experimental.
     :param corr_threshold: Minimal correlation threshold to consider two features correlated.
     :param method: One of `pearson` (default) or `spearman`.
 

fsspark/pipeline/fs_corr_rf.py

@@ -11,7 +11,7 @@
 from fsspark.fs.ml import cv_rf_classification, get_accuracy, get_predictions, get_feature_scores
 from fsspark.fs.multivariate import multivariate_filter
 from fsspark.fs.univariate import univariate_filter
-from fsspark.fs.utils import filter_missingness_rate, impute_missing
+from fsspark.fs.utils import remove_features_by_missingness_rate, impute_missing
 from fsspark.utils.datasets import get_tnbc_data_path
 from fsspark.utils.io import import_table_as_psdf
 
@@ -26,7 +26,7 @@
 # Step 1. Data pre-processing.
 
 # a) Filter missingness rate
-fsdf = filter_missingness_rate(fsdf, threshold=0.1)
+fsdf = remove_features_by_missingness_rate(fsdf, threshold=0.1)
 
 # b) Impute data frame
 fsdf = impute_missing(fsdf)