Add mode

lib/explorer/backend/lazy_series.ex

@@ -72,6 +72,7 @@ defmodule Explorer.Backend.LazySeries do
     max: 1,
     mean: 1,
     median: 1,
+    mode: 1,
     n_distinct: 1,
     variance: 1,
     standard_deviation: 1,
@@ -99,6 +100,7 @@ defmodule Explorer.Backend.LazySeries do
     :max,
     :mean,
     :median,
+    :mode,
     :variance,
     :standard_deviation,
     :count,

lib/explorer/backend/series.ex

@@ -53,6 +53,7 @@ defmodule Explorer.Backend.Series do
   @callback max(s) :: number() | Date.t() | NaiveDateTime.t() | lazy_s() | nil
   @callback mean(s) :: float() | lazy_s() | nil
   @callback median(s) :: float() | lazy_s() | nil
+  @callback mode(s) :: number() | lazy_s() | nil
   @callback variance(s) :: float() | lazy_s() | nil
   @callback standard_deviation(s) :: float() | lazy_s() | nil
   @callback quantile(s, float()) :: number | Date.t() | NaiveDateTime.t() | lazy_s() | nil

lib/explorer/polars_backend/expression.ex

@@ -40,6 +40,7 @@ defmodule Explorer.PolarsBackend.Expression do
     mean: 1,
     median: 1,
     min: 1,
+    mode: 1,
     multiply: 2,
     n_distinct: 1,
     nil_count: 1,

lib/explorer/polars_backend/series.ex

@@ -148,6 +148,9 @@ defmodule Explorer.PolarsBackend.Series do
   @impl true
   def median(series), do: Shared.apply_series(series, :s_median)
 
+  @impl true
+  def mode(series), do: Shared.apply_series(series, :s_mode)
+
   @impl true
   def variance(series), do: Shared.apply_series(series, :s_variance)
 

lib/explorer/series.ex

@@ -1241,6 +1241,27 @@ defmodule Explorer.Series do
 
   def mean(%Series{dtype: dtype}), do: dtype_error("mean/1", dtype, [:integer, :float])
 
+  @doc """
+  Gets the mode value of the series.
+
+  ## Supported dtypes
+
+    * `:integer`
+    * `:float`
+
+  ## Examples
+
+      iex> s = Explorer.Series.from_list([1, 2, 2, 3])
+      iex> Explorer.Series.mode(s)
+      2
+  """
+  @doc type: :aggregation
+  @spec mode(series :: Series.t()) :: float() | nil
+  def mode(%Series{dtype: dtype} = series) when numeric_dtype?(dtype),
+    do: Shared.apply_impl(series, :mode)
+
+  def mode(%Series{dtype: dtype}), do: dtype_error("mode/1", dtype, [:integer, :float])
+
   @doc """
   Gets the median value of the series.
 

native/explorer/Cargo.toml

@@ -54,6 +54,7 @@ features = [
   "to_dummies",
   "is_in",
   "strings",
+  "mode",
 ]
 
 [dependencies.polars-ops]

native/explorer/src/expressions.rs

@@ -598,3 +598,10 @@ pub fn expr_trim_trailing(expr: ExExpr) -> ExExpr {
     let expr: Expr = expr.resource.0.clone();
     ExExpr::new(expr.str().rstrip(None))
 }
+
+#[rustler::nif]
+pub fn expr_mode(expr: ExExpr) -> ExExpr {
+    let expr: Expr = expr.resource.0.clone();
+
+    ExExpr::new(expr.mode())
+}

native/explorer/src/lib.rs

@@ -181,6 +181,7 @@ rustler::init!(
         expr_mean,
         expr_median,
         expr_min,
+        expr_mode,
         expr_n_distinct,
         expr_nil_count,
         expr_quantile,
@@ -258,6 +259,7 @@ rustler::init!(
         s_mean,
         s_median,
         s_min,
+        s_mode,
         s_multiply,
         s_n_distinct,
         s_name,

native/explorer/src/series.rs

@@ -879,3 +879,12 @@ pub fn s_trim_trailing(data: ExSeries) -> Result<ExSeries, ExplorerError> {
     // There are no eager strip functions.
     Ok(ExSeries::new(s1.utf8()?.replace(r#"[ \s]+$"#, "")?.into()))
 }
+
+#[rustler::nif(schedule = "DirtyCpu")]
+pub fn s_mode(env: Env, data: ExSeries) -> Result<Term, ExplorerError> {
+    let s = &data.resource.0;
+    match s.dtype() {
+        DataType::UInt32 | DataType::Int64 | DataType::Float64 => Ok(s.mode()?.encode(env)),
+        dt => panic!("mode/1 not implemented for {:?}", dt),
+    }
+}