Struct DataReader 

pub struct DataReader { /* private fields */ }

Reads a delimited text file into a Dataset.

The default format is one instance per line, whitespace separated, the label in column 0, # starting a comment, and no header row.

Labels must be non-negative integers. They are treated as a dense encoding, so num_labels is max_label + 1 and a file whose labels are {0, 2} simply has an empty class 1.

Implementations

impl DataReader

pub fn new() -> Self

A reader with the default format.

pub fn with_format(self, format: DataFormat) -> Self

Sets the column delimiter.

pub fn with_headers(self, has_headers: bool) -> Self

Whether the first data line is a header to skip.

pub fn with_comment_char(self, comment_char: Option<char>) -> Self

Lines starting with this character are ignored.

pub fn with_label_column(self, label_column: usize) -> Self

Which column holds the label. Defaults to 0.

pub fn auto_detect_format(self, path: &Path) -> Self

Picks the delimiter from the file extension (.csv, .tsv, otherwise whitespace).

pub fn read_file(&self, path: &Path) -> Result<Dataset, DataReaderError>

Reads a file. The returned dataset is sorted and indexed, ready to fit.

pub fn read<R: BufRead>(&self, source: R) -> Result<Dataset, DataReaderError>

Reads from any line source. read_file is this over a file.

Trait Implementations

impl Default for DataReader

fn default() -> Self

Returns the “default value” for a type.