sklears_compose/plugin_architecture/
exampleregressor_traits.rs

1//! # ExampleRegressor - Trait Implementations
2//!
3//! This module contains trait implementations for `ExampleRegressor`.
4//!
5//! ## Implemented Traits
6//!
7//! - `PluginComponent`
8//! - `PluginEstimator`
9//!
10//! 🤖 Generated with [SplitRS](https://github.com/cool-japan/splitrs)
11
12use scirs2_core::ndarray::{Array1, Array2, ArrayView1, ArrayView2};
13use sklears_core::{
14    error::{Result as SklResult, SklearsError},
15    traits::Estimator,
16    types::Float,
17};
18use std::any::Any;
19
20use super::functions::{PluginComponent, PluginEstimator};
21use super::types::{ComponentConfig, ComponentContext, ExampleRegressor};
22
23impl PluginComponent for ExampleRegressor {
24    fn component_type(&self) -> &'static str {
25        "example_regressor"
26    }
27    fn config(&self) -> &ComponentConfig {
28        &self.config
29    }
30    fn initialize(&mut self, _context: &ComponentContext) -> SklResult<()> {
31        Ok(())
32    }
33    fn clone_component(&self) -> Box<dyn PluginComponent> {
34        Box::new(self.clone())
35    }
36    fn as_any(&self) -> &dyn Any {
37        self
38    }
39    fn as_any_mut(&mut self) -> &mut dyn Any {
40        self
41    }
42}
43
44impl PluginEstimator for ExampleRegressor {
45    fn fit(&mut self, x: &ArrayView2<'_, Float>, y: &ArrayView1<'_, Float>) -> SklResult<()> {
46        let n_features = x.ncols();
47        let mut coefficients = Array1::zeros(n_features);
48        let y_f64 = y.mapv(|v| v);
49        for _ in 0..1000 {
50            let predictions = x.dot(&coefficients);
51            let errors = &predictions - &y_f64;
52            let gradient = x.t().dot(&errors) / x.nrows() as f64;
53            coefficients = coefficients - self.learning_rate * gradient;
54        }
55        self.coefficients = Some(coefficients);
56        self.fitted = true;
57        Ok(())
58    }
59    fn predict(&self, x: &ArrayView2<'_, Float>) -> SklResult<Array1<f64>> {
60        if !self.fitted {
61            return Err(SklearsError::InvalidOperation(
62                "Estimator must be fitted before predict".to_string(),
63            ));
64        }
65        let coefficients = self.coefficients.as_ref().ok_or_else(|| {
66            SklearsError::InvalidOperation("Coefficients not initialized".to_string())
67        })?;
68        Ok(x.dot(coefficients))
69    }
70    fn score(&self, x: &ArrayView2<'_, Float>, y: &ArrayView1<'_, Float>) -> SklResult<f64> {
71        let predictions = self.predict(x)?;
72        let y_f64 = y.mapv(|v: Float| v);
73        let y_mean = y_f64.mean().ok_or_else(|| {
74            SklearsError::InvalidOperation("Cannot compute mean of empty array".to_string())
75        })?;
76        let ss_res = (&predictions - &y_f64).mapv(|x: f64| x.powi(2)).sum();
77        let ss_tot = y_f64.mapv(|x: f64| (x - y_mean).powi(2)).sum();
78        Ok(1.0 - ss_res / ss_tot)
79    }
80    fn is_fitted(&self) -> bool {
81        self.fitted
82    }
83    fn feature_importances(&self) -> Option<Array1<f64>> {
84        self.coefficients
85            .as_ref()
86            .map(|coefs: &Array1<f64>| coefs.mapv(f64::abs))
87    }
88}
sklears_compose/plugin_architecture/exampleregressor_traits.rs

sklears_compose/plugin_architecture/
exampleregressor_traits.rs
