Combining iloc and loc

Question

I am trying to combine iloc and loc, is there possibility?

Specifically I would like to:

• give integer in the form (:train_size) for the rows values

• give a list with the column names for the column values (replacing [0,1] in the code below)

  training_set = dataset.iloc[:train_size,[0,1]].values

trying

training_set = dataset.loc[:train_size,[list_input_and_y_parameters]].values

gives error message

TypeError: cannot do slice indexing on with these indexers [4275] of

Is there a way to do that?

Many thanks

Answer 1

You can chain this operation or use only iloc with Index.get_indexer for positions of columns in list:

training_set = dataset.iloc[:train_size].loc[:, ['col1','col2']].values

---

training_set = dataset.iloc[:train_size, df.columns.get_indexer(['col1','col2'])].values

Answer 2

As chaining loc and iloc can cause SettingWithCopyWarning, an option without a need to use Index.get_indexer could be (assuming there are no duplicates in the index):

training_set = dataset.loc[dataset.index[:train_size], ['col1','col2']].values