mkubasz/duck_rs

Machine Learning Library

Duck ML

Duck is implementation in Rust Machine Learning Library.

In this moment we start creating DataFrame. We want to use ndarray as some numpy alternative and nalgebra as mathematic part of coding.

Contributors:

Mateusz Kubaszek - mkubasz
Paweł Walus - Losseheil

Ducky's Structures:

type DataFrame -> Data Enginnering
    eg: 
    DataFrame::new(
                vec![
                    row![0.4, 0.7, "poster", true, 1],
                    row![3.0, 4.7, "table", true, 1],
                    row![3.0, 4.7, "book", true, 1],
                ],
                vec!["A", "B", "C", "D", "E"]
            );
type DataFrameGroupBy -> Hashmap
type Series -> Vectorize structure
    eg:
    Series::new(
        row![0.4, 0.7, "poster", true, 1],
        String::from("A")
    )
type Cell -> One element in Data Frame
    - Number
    - Float
    - Float64
    - Bool
    - Text
type Matrix -> Algebraic matrix

Ducky's Operations:

fn push(&mut self, element: Vec<Cell>);

fn series(&mut self, index: usize) -> &mut Series<Cell>;

fn to_rows(&self) -> Option<Vec<Vec<Cell>>>;

fn by(&mut self, label: &str) -> Option<&mut Series<Cell>>;

fn many(&mut self, labels: Vec<&str>) -> Vec<Series<Cell>>;

fn map(&mut self, col: &str, obj: HashMap<&str, u32>) -> DataFrame;

fn concat(&mut self, df: DataFrame) -> DataFrame;

fn drop(&mut self, label: Vec<&str>) -> Option<DataFrame>;

fn drop_idx(&mut self, position: usize) -> Option<DataFrame>;

fn contains(self, label: &str) -> bool;

fn join(&mut self, df: DataFrame);

fn group_by(&mut self, label: &str) -> Option<DataFrameGroupBy>;

fn sort(&mut self, label: &str) -> Option<DataFrame>;

fn read_csv(file_name: String) -> Result<DataFrame, Box<dyn Error>>;

Ducky's Science:

fn get_dummies(&mut self, label: &str) -> DataFrame;

fn from_vec<T>(vec: Vec<Vec<T>>, labels: Vec<&str>) -> DataFrame where T: PrimInt + Signed + Product;

fn to_matrix(&self) -> Option<Matrix>;

Ducky's Math:

fn sigmoid(x: f64) -> f64;

fn minmax_item<T: Into<f64> + Copy>(min_max: (f64, f64), item: T) -> (f64, f64);

fn minmax<T: Into<f64> + Copy>(series: Vec<T>) -> Option<(f64, f64)>;

fn series_minmax(series: Series<Cell>) -> (f64, f64);

Development milestone:

In first step:

• Algorithms: Evaluation Methods, Evaluation Metrics, Baseline Models.

In second step:

• Algorithms: (Simple/Multivariate) Linear Regression, Logistic Regression, Perceptron.

In third step:

• Algorithms: Decision Trees, Regression Trees, Naive Bayes, k-nearest, Backpropagation.