dataset_loaders/
dataset_loaders.rs

1use scirs2_datasets::loaders;
2use scirs2_datasets::utils::{train_test_split, Dataset};
3use std::env;
4use std::path::Path;
5
6#[allow(dead_code)]
7fn main() {
8    // Check if a CSV file is provided as a command-line argument
9    let args: Vec<String> = env::args().collect();
10    if args.len() < 2 {
11        println!("Usage: {} <path_to_csv_file>", args[0]);
12        println!("Example: {} examples/sampledata.csv", args[0]);
13        return;
14    }
15
16    let filepath = &args[1];
17
18    // Verify the file exists
19    if !Path::new(filepath).exists() {
20        println!("Error: File '{filepath}' does not exist");
21        return;
22    }
23
24    // Load CSV file
25    println!("Loading CSV file: {filepath}");
26    let csv_config = loaders::CsvConfig {
27        has_header: true,
28        target_column: None,
29        ..Default::default()
30    };
31    match loaders::load_csv(filepath, csv_config) {
32        Ok(dataset) => {
33            print_dataset_info(&dataset, "Loaded CSV");
34
35            // Split the dataset for demonstration
36            println!("\nDemonstrating train-test split...");
37            match train_test_split(&dataset, 0.2, Some(42)) {
38                Ok((train, test)) => {
39                    println!("Training set: {} samples", train.n_samples());
40                    println!("Test set: {} samples", test.n_samples());
41
42                    // Save as JSON for demonstration
43                    let jsonpath = format!("{filepath}.json");
44                    println!("\nSaving training dataset to JSON: {jsonpath}");
45                    if let Err(e) = loaders::save_json(&train, &jsonpath) {
46                        println!("Error saving JSON: {e}");
47                    } else {
48                        println!("Successfully saved JSON file");
49
50                        // Load back the JSON file
51                        println!("\nLoading back from JSON file...");
52                        match loaders::load_json(&jsonpath) {
53                            Ok(loaded) => {
54                                print_dataset_info(&loaded, "Loaded JSON");
55                            }
56                            Err(e) => println!("Error loading JSON: {e}"),
57                        }
58                    }
59                }
60                Err(e) => println!("Error splitting dataset: {e}"),
61            }
62        }
63        Err(e) => println!("Error loading CSV: {e}"),
64    }
65}
66
67#[allow(dead_code)]
68fn print_dataset_info(dataset: &Dataset, name: &str) {
69    println!("=== {name} Dataset ===");
70    println!("Number of samples: {}", dataset.n_samples());
71    println!("Number of features: {}", dataset.n_features());
72
73    if let Some(featurenames) = &dataset.featurenames {
74        println!(
75            "Features: {:?}",
76            &featurenames[0..std::cmp::min(5, featurenames.len())]
77        );
78        if featurenames.len() > 5 {
79            println!("... and {} more", featurenames.len() - 5);
80        }
81    }
82
83    if let Some(target) = &dataset.target {
84        println!("Target shape: {}", target.len());
85
86        if let Some(targetnames) = &dataset.targetnames {
87            println!("Target classes: {targetnames:?}");
88        }
89    }
90
91    for (key, value) in &dataset.metadata {
92        println!("Metadata - {key}: {value}");
93    }
94}
dataset_loaders/dataset_loaders.rs

dataset_loaders/
dataset_loaders.rs
