Lint & small changes
This commit is contained in:
parent
0c6d207967
commit
55238f59aa
21 changed files with 2522 additions and 423 deletions
|
|
@ -1,6 +1,13 @@
|
|||
use crate::consts::NAN_U16;
|
||||
use crate::data::QuantRef;
|
||||
|
||||
/// Optional per-enum-value distribution tracking for a single feature.
|
||||
/// Counts how many rows have each enum value (by raw u16 index).
|
||||
pub struct EnumDist {
|
||||
pub feat_idx: usize,
|
||||
pub counts: Box<[u32]>,
|
||||
}
|
||||
|
||||
/// Per-cell accumulator for aggregating features (min/max/sum/count).
|
||||
/// Uses Box<[T]> instead of Vec<T> to avoid storing capacity (saves 8 bytes per field per cell).
|
||||
/// Shared by hexagon and postcode aggregation routes.
|
||||
|
|
@ -10,16 +17,26 @@ pub struct Aggregator {
|
|||
pub maxs: Box<[f32]>,
|
||||
pub sums: Box<[f64]>,
|
||||
pub feat_counts: Box<[u32]>,
|
||||
/// Optional: per-value counts for a single enum feature (for pie chart visualization).
|
||||
pub enum_dist: Option<EnumDist>,
|
||||
}
|
||||
|
||||
/// Configuration for enum distribution tracking, passed to Aggregator::new.
|
||||
/// (feature_index, number_of_enum_values)
|
||||
pub type EnumDistConfig = Option<(usize, usize)>;
|
||||
|
||||
impl Aggregator {
|
||||
pub fn new(num_features: usize) -> Self {
|
||||
pub fn new(num_features: usize, enum_dist_config: EnumDistConfig) -> Self {
|
||||
Aggregator {
|
||||
count: 0,
|
||||
mins: vec![f32::INFINITY; num_features].into_boxed_slice(),
|
||||
maxs: vec![f32::NEG_INFINITY; num_features].into_boxed_slice(),
|
||||
sums: vec![0.0f64; num_features].into_boxed_slice(),
|
||||
feat_counts: vec![0u32; num_features].into_boxed_slice(),
|
||||
enum_dist: enum_dist_config.map(|(feat_idx, num_values)| EnumDist {
|
||||
feat_idx,
|
||||
counts: vec![0u32; num_values].into_boxed_slice(),
|
||||
}),
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -50,6 +67,17 @@ impl Aggregator {
|
|||
self.feat_counts[feat_index] += 1;
|
||||
}
|
||||
}
|
||||
// Enum distribution: single branch per row (not per feature).
|
||||
// Uses raw u16 directly — enum features are stored as u16 indices.
|
||||
if let Some(ref mut ed) = self.enum_dist {
|
||||
let raw = row_slice[ed.feat_idx];
|
||||
if raw != NAN_U16 {
|
||||
let idx = raw as usize;
|
||||
if idx < ed.counts.len() {
|
||||
ed.counts[idx] += 1;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Merge another aggregator's results into this one.
|
||||
|
|
@ -67,6 +95,12 @@ impl Aggregator {
|
|||
self.feat_counts[i] += other.feat_counts[i];
|
||||
}
|
||||
}
|
||||
// Merge enum distribution counts
|
||||
if let (Some(ref mut mine), Some(ref theirs)) = (&mut self.enum_dist, &other.enum_dist) {
|
||||
for (m, t) in mine.counts.iter_mut().zip(theirs.counts.iter()) {
|
||||
*m += t;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Add a row, only aggregating the features at the given indices.
|
||||
|
|
@ -95,5 +129,15 @@ impl Aggregator {
|
|||
self.feat_counts[feat_index] += 1;
|
||||
}
|
||||
}
|
||||
// Enum distribution (same raw u16 approach)
|
||||
if let Some(ref mut ed) = self.enum_dist {
|
||||
let raw = feature_data[base + ed.feat_idx];
|
||||
if raw != NAN_U16 {
|
||||
let idx = raw as usize;
|
||||
if idx < ed.counts.len() {
|
||||
ed.counts[idx] += 1;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue