sklearn-rs 0.1.0

use ndarray::{Array1, Array2};
use crate::error::{Result, SklearnError};

/// 类似于 scikit-learn 的 Estimator trait
pub trait Estimator {
    type Input;
    type Target;
    type Model: Predictor<Input = Self::Input>;
    
    fn fit(&self, x: &Self::Input, y: &Self::Target) -> Result<Self::Model>;
}

/// 预测器 trait
pub trait Predictor {
    type Input;
    type Output;
    
    fn predict(&self, x: &Self::Input) -> Result<Self::Output>;
}

/// 数据转换器 trait  
pub trait Transformer {
    type Input;
    type Output;
    
    fn fit(&self, x: &Self::Input) -> Result<Self>
    where
        Self: Sized;
        
    fn transform(&self, x: &Self::Input) -> Result<Self::Output>;
    
    fn fit_transform(&mut self, x: &Self::Input) -> Result<Self::Output>
    where
        Self: Sized,
    {
        let _ = self.fit(x)?;
        self.transform(x)
    }
}

/// 验证输入数据的基本函数
pub fn validate_features(x: &Array2<f64>) -> Result<()> {
    if x.nrows() == 0 {
        return Err(SklearnError::ShapeMismatch {
            expected: "至少一行".to_string(),
            actual: "0行".to_string(),
        });
    }
    
    if x.ncols() == 0 {
        return Err(SklearnError::ShapeMismatch {
            expected: "至少一列".to_string(),
            actual: "0列".to_string(),
        });
    }
    
    for &val in x.iter() {
        if val.is_nan() || val.is_infinite() {
            return Err(SklearnError::InvalidData);
        }
    }
    
    Ok(())
}

/// 验证目标值
pub fn validate_target(y: &Array1<f64>) -> Result<()> {
    if y.is_empty() {
        return Err(SklearnError::ShapeMismatch {
            expected: "非空数组".to_string(),
            actual: "空数组".to_string(),
        });
    }
    
    for &val in y.iter() {
        if val.is_nan() || val.is_infinite() {
            return Err(SklearnError::InvalidData);
        }
    }
    
    Ok(())
}