reorganize pca-analysis

collector/src/mir_analyze/data/table_data.rs

@@ -239,7 +239,7 @@ impl<X: Debug + Clone + Ord, Y: Debug + Clone + Ord + Display, T: Debug + Copy +
     }
 }
 
-fn into_vec<X: Clone + Ord, Y: Clone + Ord, T: Copy + Into<f64>>(
+pub fn into_vec<X: Clone + Ord, Y: Clone + Ord, T: Copy + Into<f64>>(
     table_data: &TableDatas<X, Y, T>,
 ) -> Vec<f64> {
     let mut data = vec![];
@@ -252,7 +252,7 @@ fn into_vec<X: Clone + Ord, Y: Clone + Ord, T: Copy + Into<f64>>(
 }
 
 /// Sort TableData by axis X and Y.
-fn sort<X: Ord + Clone, Y: Ord + Clone, T: Clone>(
+pub fn sort<X: Ord + Clone, Y: Ord + Clone, T: Clone>(
     table_data: &TableDatas<X, Y, T>,
 ) -> Vec<(X, Vec<(Y, T)>)> {
     let mut data_sorted = table_data

collector/src/pca_analysis/plotter/coordinate_map.rs

@@ -95,11 +95,9 @@ fn draw(
 mod coordinate_map_test {
     use std::{collections::HashMap, fs, path::PathBuf};
 
-    use nalgebra::DVector;
-
     use crate::mir_analyze::data::table_data::{generate_benchmark_data, TableDatas};
 
-    use super::{draw_coordinate_map_2d, get_coordinate_2d, get_principle_components};
+    use super::{draw_coordinate_map_2d, get_principle_components};
 
     fn generate_table_data() -> TableDatas<String, String, i32> {
         vec![
@@ -149,41 +147,6 @@ mod coordinate_map_test {
         .collect::<HashMap<String, HashMap<String, i32>>>()
     }
 
-    /// A test for `get_coordinate_2d`.
-    ///
-    /// Step1. Create datas and features.
-    ///
-    /// Step2. Calculate coordinate of each data.
-    ///
-    /// Step3. Verify the result.
-    #[test]
-    fn test_get_coordinate_2d() {
-        let datas = vec![
-            vec![1.0, 4.0, 7.0],
-            vec![2.0, 5.0, 8.0],
-            vec![3.0, 6.0, 9.0],
-        ];
-        let feature_x = DVector::from_vec(vec![
-            0.8164965809277261,
-            -0.4082482904638633,
-            -0.408248290463863,
-        ]);
-        let feature_y = DVector::from_vec(vec![0.0, -0.7071067811865475, 0.7071067811865477]);
-        let mut results = vec![
-            (-3.674234614174768, 2.1213203435596437),
-            (-3.674234614174768, 2.121320343559644),
-            (-3.6742346141747686, 2.1213203435596446),
-        ]
-        .into_iter();
-
-        datas.into_iter().for_each(|d| {
-            assert_eq!(
-                get_coordinate_2d(&DVector::from(d), &feature_x, &feature_y),
-                results.next().unwrap()
-            );
-        });
-    }
-
     /// A test for `draw_coordinate_map_2d`.
     ///
     /// Step1. Draw PCA charts.

data_manage/Cargo.toml

@@ -9,5 +9,6 @@ edition = "2021"
 anyhow = "1"
 collector = { path = "../collector" }
 clap = { version = "3.0.9", features = ["derive"] }
+nalgebra = "0.32.4"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"

data_manage/src/commannds.rs

@@ -97,9 +97,9 @@ pub enum Commands {
         /// The path of output file.
         #[clap(long = "out-path")]
         out_path: PathBuf,
-        /// Metrics need to be merged.
+        /// Metrics need to be merged. Use ',' to concanate the metrics.
         #[clap(long = "old-metrics")]
-        old_metrics: Vec<String>,
+        old_metrics: String,
         /// New metric merged from old-metrics.
         #[clap(long = "merged-metric")]
         merged_metric: String,
@@ -116,8 +116,21 @@ pub enum Commands {
         /// The path of output file.
         #[clap(long = "out-path")]
         out_path: PathBuf,
-        /// Metrics merged from stats fmt file.
+        /// Metrics merged from stats fmt file. Use ',' to concanate the metrics.
         #[clap(long = "new_metrics")]
-        new_metrics: Vec<String>,
+        new_metrics: String,
+    },
+
+    /// Do pca analysis on a table fmt file.
+    PcaAnalysis {
+        /// The path of table data fmt file.
+        #[clap(long = "table-data")]
+        table_data_path: PathBuf,
+        /// The path of output dir.
+        #[clap(long = "out-dir")]
+        out_dir: PathBuf,
+        /// The maximun number principle components.
+        #[clap(long = "max-component-num")]
+        max_component_num: u32,
     },
 }

data_manage/src/main.rs

@@ -2,13 +2,15 @@ use clap::Parser;
 use commannds::Cli;
 use compare_stats::{compare_stat::compare_stat, compare_stat_2d::compare_stat_2d};
 use merge_stats::{merge_runtime_stat::merge_runtime_stats, merge_stat::merge_compile_time_stats};
+use pca_analysis::entry::pca_entry;
 use table_data::merge_metrics::{
     merge_metrics_from_compile_time_stats_to_table_data, merge_metrics_on_table_data,
 };
 
 mod commannds;
 mod compare_stats;
 mod merge_stats;
+mod pca_analysis;
 mod table_data;
 
 fn main() {
@@ -59,7 +61,7 @@ fn main() {
         } => match merge_metrics_on_table_data(
             &table_data_path,
             &out_path,
-            &old_metrics,
+            &old_metrics.split(',').map(|s| s.to_string()).collect(),
             &merged_metric,
         ) {
             Ok(p) => println!("Write merged table data to {}", p.to_str().unwrap()),
@@ -74,10 +76,18 @@ fn main() {
             &table_data_path,
             &stats_path,
             out_path.as_path(),
-            new_metrics,
+            new_metrics.split(',').map(|s| s.to_string()).collect(),
         ) {
             Ok(p) => println!("Write merged table data to {}", p.to_str().unwrap()),
             Err(e) => eprintln!("{}", e),
         },
+        commannds::Commands::PcaAnalysis {
+            table_data_path,
+            out_dir,
+            max_component_num,
+        } => match pca_entry(&table_data_path, out_dir, max_component_num) {
+            Ok(p) => println!("Write Pca analysis result to {}", p.to_str().unwrap()),
+            Err(e) => eprintln!("{}", e),
+        },
     }
 }

data_manage/src/pca_analysis/entry.rs

@@ -0,0 +1,51 @@
+use std::{
+    fs::{create_dir_all, File},
+    io::BufReader,
+    path::PathBuf,
+};
+
+use collector::mir_analyze::data::table_data::{sort, TableDatas};
+
+use super::{pca_data::get_principle_components, plotter::coordinate_map::draw_coordinate_map_2d};
+
+pub fn pca_entry(
+    table_data_path: &PathBuf,
+    out_dir: PathBuf,
+    max_component_num: u32,
+) -> anyhow::Result<PathBuf> {
+    let table_data: TableDatas<String, String, f64> =
+        serde_json::from_reader(BufReader::new(File::open(table_data_path)?))?;
+
+    let pc = get_principle_components(&table_data, max_component_num);
+    display_pc(&pc, &table_data);
+
+    draw_coordinate_map_2d(&pc, &table_data, &out_dir)?;
+
+    create_dir_all(&out_dir)?;
+
+    Ok(out_dir)
+}
+
+pub(super) fn display_pc(pcs: &Vec<Vec<f64>>, table_data: &TableDatas<String, String, f64>) {
+    let labels = sort(table_data)
+        .into_iter()
+        .map(|(s, _)| s)
+        .collect::<Vec<String>>()
+        .join("\t");
+
+    let mut k = 0;
+    println!("=====================================================");
+    println!("{}", labels);
+    for pc in pcs {
+        k += 1;
+        println!(
+            "PC{}: {}",
+            k,
+            pc.into_iter()
+                .map(|p| format!("{:.2}", p))
+                .collect::<Vec<String>>()
+                .join("\t")
+        );
+    }
+    println!("=====================================================");
+}

data_manage/src/pca_analysis/mod.rs

@@ -0,0 +1,3 @@
+pub(super) mod entry;
+mod pca_data;
+mod plotter;