Version 1.0.1

mrdimosthenis · Sep 26, 2024 · 6ff7b04 · 6ff7b04
1 parent 732dd36
commit 6ff7b04
Show file tree

Hide file tree

Showing 31 changed files with 273 additions and 332 deletions.
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
@@ -11,11 +11,12 @@ jobs:
   test:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v2.0.0
-      - uses: erlef/setup-beam@v1.9.0
+      - uses: actions/checkout@v4
+      - uses: erlef/setup-beam@v1
         with:
-          otp-version: "23.2"
-          gleam-version: "0.22.1"
-      - run: gleam format --check src test
+          otp-version: "27.0.0"
+          gleam-version: "1.5.1"
+          rebar3-version: "3"
       - run: gleam deps download
       - run: gleam test
+      - run: gleam format --check src test
diff --git a/README.md b/README.md
@@ -8,7 +8,7 @@ Turn data into functions! A simple and functional **machine learning** library,
 
 ```erlang
 {deps, [
-    {emel, "1.0.0"}
+    {emel, "1.0.1"}
 ]}.
 ```
 
@@ -17,7 +17,7 @@ Turn data into functions! A simple and functional **machine learning** library,
 ```elixir
 defp deps do
   [
-    {:emel, "~> 1.0.0"}
+    {:emel, "~> 1.0.1"}
   ]
 end
 ```

diff --git a/gleam.toml b/gleam.toml
@@ -1,5 +1,5 @@
 name = "emel"
-version = "1.0.0"
+version = "1.0.1"
 licences = ["Apache-2.0"]
 description = "Turn data into functions in the Erlang ecosystem"
 
@@ -14,10 +14,10 @@ user = "mrdimosthenis"
 repo = "emel"
 
 [dependencies]
-gleam_stdlib = "~> 0.22"
-minigen = "~> 0.0"
-gleam_zlists = "~> 0.0"
-gleam_synapses = "~> 1.0"
+gleam_stdlib = "~> 0.40.0"
+minigen = "~> 0.1.1"
+gleam_zlists = "~> 1.0.0"
+gleam_synapses = "~> 1.0.1"
 
 [dev-dependencies]
-gleeunit = "~> 0.6"
+gleeunit = "~> 1.2.0"
diff --git a/manifest.toml b/manifest.toml
@@ -2,19 +2,18 @@
 # You typically do not need to edit this file
 
 packages = [
-  { name = "gleam_json", version = "0.5.0", build_tools = ["gleam"], requirements = ["gleam_stdlib", "thoas"], otp_app = "gleam_json", source = "hex", outer_checksum = "E42443C98AA66E30143C24818F2CEA801491C10CE6B1A5EDDF3FC4ABDC7601CB" },
-  { name = "gleam_stdlib", version = "0.22.1", build_tools = ["gleam"], requirements = [], otp_app = "gleam_stdlib", source = "hex", outer_checksum = "9C6C66EDAAFD097267089565488DEDFC111A0497DE9CB2C42248E26330CC48E9" },
-  { name = "gleam_synapses", version = "1.0.0", build_tools = ["gleam"], requirements = ["gleam_json", "minigen", "gleam_stdlib", "gleam_zlists"], otp_app = "gleam_synapses", source = "hex", outer_checksum = "B8D86A43EADC88D951AC9B3E40F667AFCDB30BA13485123E503041FBA542DE04" },
-  { name = "gleam_zlists", version = "0.0.4", build_tools = ["gleam"], requirements = ["gleam_stdlib", "zlists"], otp_app = "gleam_zlists", source = "hex", outer_checksum = "A71998C938A90475051845960ABD73647DB9734088E78A58B0CDFC41B1486461" },
-  { name = "gleeunit", version = "0.6.1", build_tools = ["gleam"], requirements = ["gleam_stdlib"], otp_app = "gleeunit", source = "hex", outer_checksum = "5BF486C3E135B7F5ED8C054925FC48E5B2C79016A39F416FD8CF2E860520EE55" },
-  { name = "minigen", version = "0.0.3", build_tools = ["gleam"], requirements = ["gleam_stdlib"], otp_app = "minigen", source = "hex", outer_checksum = "F74D7B2BF5E05A493BF2048BB557F8FC08AA93EB5A6664FC66FD5A69C3A57B6A" },
-  { name = "thoas", version = "0.3.0", build_tools = ["rebar3"], requirements = [], otp_app = "thoas", source = "hex", outer_checksum = "B8E1F8C8FAD317C0B75239A9234CB093DE1FB8BE7BA3E41433FF80A0B3353973" },
+  { name = "gleam_json", version = "2.0.0", build_tools = ["gleam"], requirements = ["gleam_stdlib"], otp_app = "gleam_json", source = "hex", outer_checksum = "CB10B0E7BF44282FB25162F1A24C1A025F6B93E777CCF238C4017E4EEF2CDE97" },
+  { name = "gleam_stdlib", version = "0.40.0", build_tools = ["gleam"], requirements = [], otp_app = "gleam_stdlib", source = "hex", outer_checksum = "86606B75A600BBD05E539EB59FABC6E307EEEA7B1E5865AFB6D980A93BCB2181" },
+  { name = "gleam_synapses", version = "1.0.1", build_tools = ["gleam"], requirements = ["gleam_json", "gleam_stdlib", "gleam_zlists", "minigen"], otp_app = "gleam_synapses", source = "hex", outer_checksum = "2097F516EF2131B6FCE3D2058C787632D7BFDEE3B91AFD3CBF669D43CE722BCE" },
+  { name = "gleam_zlists", version = "1.0.0", build_tools = ["gleam"], requirements = ["gleam_stdlib", "zlists"], otp_app = "gleam_zlists", source = "hex", outer_checksum = "DCA7EAAAB38C0BB3A16E7490EFF51DA0C2355A63B60BC3DD98752FDE96314B94" },
+  { name = "gleeunit", version = "1.2.0", build_tools = ["gleam"], requirements = ["gleam_stdlib"], otp_app = "gleeunit", source = "hex", outer_checksum = "F7A7228925D3EE7D0813C922E062BFD6D7E9310F0BEE585D3A42F3307E3CFD13" },
+  { name = "minigen", version = "0.1.1", build_tools = ["gleam"], requirements = ["gleam_stdlib"], otp_app = "minigen", source = "hex", outer_checksum = "AFD1E637A7F429C2C57438FB635CCCAD818AA9C39D6F6E1A5200E5952940E2FD" },
   { name = "zlists", version = "0.0.4", build_tools = ["mix", "rebar3"], requirements = [], otp_app = "zlists", source = "hex", outer_checksum = "BABC8B78984E4988DAFEB9E07A77F04D3F8EA208D49E60B6669DA43B0D2C2C4B" },
 ]
 
 [requirements]
-gleam_stdlib = "~> 0.22"
-gleam_synapses = "~> 1.0"
-gleam_zlists = "~> 0.0"
-gleeunit = "~> 0.6"
-minigen = "~> 0.0"
+gleam_stdlib = { version = "~> 0.40.0" }
+gleam_synapses = { version = "~> 1.0.1" }
+gleam_zlists = { version = "~> 1.0.0" }
+gleeunit = { version = "~> 1.2.0" }
+minigen = { version = "~> 0.1.1" }
diff --git a/src/emel/lazy/math/algebra.gleam b/src/emel/lazy/math/algebra.gleam
@@ -3,7 +3,7 @@ import emel/utils/zlist as ut_zlist
 import gleam/float
 import gleam/int
 import gleam/order.{Eq}
-import gleam_zlists.{ZList} as zlist
+import gleam_zlists.{type ZList} as zlist
 
 pub fn first_minor(
   matrix: ZList(ZList(Float)),
@@ -19,7 +19,7 @@ pub fn determinant(matrix: ZList(ZList(Float))) -> Float {
   case zlist.count(matrix) {
     0 -> 1.0
     2 -> {
-      let [[a, b], [c, d]] = ut_zlist.to_list_of_lists(matrix)
+      let assert [[a, b], [c, d]] = ut_zlist.to_list_of_lists(matrix)
       a *. d -. c *. b
     }
     _ ->

diff --git a/src/emel/lazy/math/geometry.gleam b/src/emel/lazy/math/geometry.gleam
@@ -1,8 +1,7 @@
 import emel/utils/result as ut_res
 import emel/utils/zlist as ut_zlist
 import gleam/float
-import gleam/function
-import gleam_zlists.{ZList} as zlist
+import gleam_zlists.{type ZList} as zlist
 
 pub fn dot_product(x: ZList(Float), y: ZList(Float)) -> Float {
   zlist.zip(x, y)
@@ -27,7 +26,7 @@ pub fn euclidean_distance(x: ZList(Float), y: ZList(Float)) -> Float {
 
 pub fn magnitude(vector: ZList(Float)) -> Float {
   vector
-  |> zlist.map(function.constant(0.0))
+  |> zlist.map(fn(_) { 0.0 })
   |> euclidean_distance(vector)
 }
 
@@ -46,20 +45,16 @@ pub fn centroid(points: ZList(ZList(Float))) -> ZList(Float) {
     |> zlist.uncons
     |> ut_res.unsafe_res
   let #(s, n) =
-    zlist.reduce(
-      tl,
-      #(hd, 1.0),
-      fn(v, acc) {
-        let #(acc_s, acc_n) = acc
-        let next_acc_s =
-          zlist.zip(acc_s, v)
-          |> zlist.map(fn(t) {
-            let #(si, vi) = t
-            si +. vi
-          })
-        let next_acc_n = acc_n +. 1.0
-        #(next_acc_s, next_acc_n)
-      },
-    )
+    zlist.reduce(tl, #(hd, 1.0), fn(v, acc) {
+      let #(acc_s, acc_n) = acc
+      let next_acc_s =
+        zlist.zip(acc_s, v)
+        |> zlist.map(fn(t) {
+          let #(si, vi) = t
+          si +. vi
+        })
+      let next_acc_n = acc_n +. 1.0
+      #(next_acc_s, next_acc_n)
+    })
   zlist.map(s, fn(x) { x /. n })
 }
diff --git a/src/emel/lazy/math/statistics.gleam b/src/emel/lazy/math/statistics.gleam
@@ -1,9 +1,9 @@
 import gleam/float
 import gleam/int
-import gleam_zlists.{ZList} as zlist
+import gleam_zlists.{type ZList} as zlist
 
-external fn log(Float) -> Float =
-  "math" "log"
+@external(erlang, "math", "log")
+pub fn log(x: Float) -> Float
 
 fn log_base(x: Float, b: Float) -> Float {
   log(x) /. log(b)
@@ -21,19 +21,15 @@ pub fn entropy(probability_values: ZList(Float)) -> Float {
 
 pub fn classical_probability(observations: ZList(a), event: a) -> Float {
   let #(numerator, denominator) =
-    zlist.reduce(
-      observations,
-      #(0.0, 0.0),
-      fn(ev, acc) {
-        let #(numer, denom) = acc
-        let next_numer = case ev == event {
-          True -> numer +. 1.0
-          False -> numer
-        }
-        let next_denom = denom +. 1.0
-        #(next_numer, next_denom)
-      },
-    )
+    zlist.reduce(observations, #(0.0, 0.0), fn(ev, acc) {
+      let #(numer, denom) = acc
+      let next_numer = case ev == event {
+        True -> numer +. 1.0
+        False -> numer
+      }
+      let next_denom = denom +. 1.0
+      #(next_numer, next_denom)
+    })
   numerator /. denominator
 }
 
@@ -42,16 +38,12 @@ pub fn mean_absolute_error(
   observations: ZList(Float),
 ) -> Float {
   let #(numerator, denominator) =
-    zlist.reduce(
-      zlist.zip(predictions, observations),
-      #(0.0, 0.0),
-      fn(p, acc) {
-        let #(v1, v2) = p
-        let #(numer, denom) = acc
-        let next_numer = numer +. float.absolute_value(v1 -. v2)
-        let next_denom = denom +. 1.0
-        #(next_numer, next_denom)
-      },
-    )
+    zlist.reduce(zlist.zip(predictions, observations), #(0.0, 0.0), fn(p, acc) {
+      let #(v1, v2) = p
+      let #(numer, denom) = acc
+      let next_numer = numer +. float.absolute_value(v1 -. v2)
+      let next_denom = denom +. 1.0
+      #(next_numer, next_denom)
+    })
   numerator /. denominator
 }
diff --git a/src/emel/lazy/ml/decision_tree.gleam b/src/emel/lazy/ml/decision_tree.gleam
@@ -1,13 +1,13 @@
 import emel/lazy/math/statistics as stats
 import emel/utils/result as ut_res
 import emel/utils/zlist as ut_zlist
+import gleam/dict.{type Dict}
 import gleam/int
-import gleam/map.{Map}
 import gleam/pair
-import gleam_zlists.{ZList} as zlist
+import gleam_zlists.{type ZList} as zlist
 
 pub fn entropy_with_size(
-  dataset: ZList(Map(String, String)),
+  dataset: ZList(Dict(String, String)),
   class_attr: String,
 ) -> #(Float, Int) {
   let #(freqs, size) =
@@ -23,7 +23,7 @@ pub fn entropy_with_size(
 }
 
 pub fn feature_entropy(
-  dataset: ZList(Map(String, String)),
+  dataset: ZList(Dict(String, String)),
   class_attr: String,
   feature: String,
   dataset_size: Int,
@@ -40,23 +40,23 @@ pub fn feature_entropy(
 }
 
 fn same_class(
-  rule: Map(String, String),
+  rule: Dict(String, String),
   class_attr: String,
-  sub_dataset: ZList(Map(String, String)),
-) -> ZList(Map(String, String)) {
+  sub_dataset: ZList(Dict(String, String)),
+) -> ZList(Dict(String, String)) {
   sub_dataset
   |> zlist.head
   |> ut_res.unsafe_res
   |> ut_res.unsafe_get(class_attr)
-  |> map.insert(rule, class_attr, _)
+  |> dict.insert(rule, class_attr, _)
   |> zlist.singleton
 }
 
 fn exhausted_attributes(
-  rule: Map(String, String),
+  rule: Dict(String, String),
   class_attr: String,
-  grouped_by_class: ZList(#(String, ZList(Map(String, String)))),
-) -> ZList(Map(String, String)) {
+  grouped_by_class: ZList(#(String, ZList(Dict(String, String)))),
+) -> ZList(Dict(String, String)) {
   grouped_by_class
   |> ut_zlist.max_by(fn(t) {
     t
@@ -66,16 +66,16 @@ fn exhausted_attributes(
   })
   |> ut_res.unsafe_res
   |> pair.first
-  |> map.insert(rule, class_attr, _)
+  |> dict.insert(rule, class_attr, _)
   |> zlist.singleton
 }
 
 fn unfold_rule(
-  rule: Map(String, String),
+  rule: Dict(String, String),
   non_selected_attrs: ZList(String),
   class_attr: String,
-  sub_dataset: ZList(Map(String, String)),
-) -> ZList(Map(String, String)) {
+  sub_dataset: ZList(Dict(String, String)),
+) -> ZList(Dict(String, String)) {
   let grouped_by_class =
     ut_zlist.group_by(sub_dataset, ut_res.unsafe_get(_, class_attr))
   case zlist.count(grouped_by_class) {
@@ -89,24 +89,19 @@ fn unfold_rule(
           let next_selected_attr =
             non_selected_attrs
             |> ut_zlist.max_by(fn(feature) {
-              entropy -. feature_entropy(
-                sub_dataset,
-                class_attr,
-                feature,
-                dataset_size,
-              )
+              entropy
+              -. feature_entropy(sub_dataset, class_attr, feature, dataset_size)
             })
             |> ut_res.unsafe_res
           let next_non_selected_attrs =
-            zlist.filter(
-              non_selected_attrs,
-              fn(attr) { attr != next_selected_attr },
-            )
+            zlist.filter(non_selected_attrs, fn(attr) {
+              attr != next_selected_attr
+            })
           sub_dataset
           |> ut_zlist.group_by(ut_res.unsafe_get(_, next_selected_attr))
           |> zlist.flat_map(fn(t) {
             let #(feature_val, sub_group) = t
-            let next_rule = map.insert(rule, next_selected_attr, feature_val)
+            let next_rule = dict.insert(rule, next_selected_attr, feature_val)
             unfold_rule(
               next_rule,
               next_non_selected_attrs,
@@ -120,32 +115,29 @@ fn unfold_rule(
 }
 
 pub fn decision_tree(
-  dataset: ZList(Map(String, String)),
+  dataset: ZList(Dict(String, String)),
   attributes: ZList(String),
   class: String,
-) -> ZList(Map(String, String)) {
-  unfold_rule(map.new(), attributes, class, dataset)
+) -> ZList(Dict(String, String)) {
+  unfold_rule(dict.new(), attributes, class, dataset)
 }
 
 pub fn classifier(
-  dataset: ZList(Map(String, String)),
+  dataset: ZList(Dict(String, String)),
   discrete_attributes: ZList(String),
   class: String,
-) -> fn(Map(String, String)) -> String {
-  let all_rules: ZList(Map(String, String)) =
+) -> fn(Dict(String, String)) -> String {
+  let all_rules: ZList(Dict(String, String)) =
     decision_tree(dataset, discrete_attributes, class)
   fn(item) {
     all_rules
     |> zlist.find(fn(rule) {
-      zlist.all(
-        discrete_attributes,
-        fn(feature) {
-          case map.get(rule, feature) {
-            Error(Nil) -> True
-            Ok(v) -> v == ut_res.unsafe_get(item, feature)
-          }
-        },
-      )
+      zlist.all(discrete_attributes, fn(feature) {
+        case dict.get(rule, feature) {
+          Error(Nil) -> True
+          Ok(v) -> v == ut_res.unsafe_get(item, feature)
+        }
+      })
     })
     |> ut_res.unsafe_res
     |> ut_res.unsafe_get(class)

diff --git a/src/emel/lazy/ml/k_means.gleam b/src/emel/lazy/ml/k_means.gleam
@@ -1,9 +1,9 @@
+import emel/lazy/math/geometry
 import emel/utils/result as ut_res
 import emel/utils/zlist as ut_zlist
-import emel/lazy/math/geometry
 import gleam/pair
-import gleam/set.{Set}
-import gleam_zlists.{ZList} as zlist
+import gleam/set.{type Set}
+import gleam_zlists.{type ZList} as zlist
 
 fn point_groups(
   points: ZList(ZList(Float)),