OrgLion-ML/golearn
1
0
Fork 0
mirror of https://github.com/sjwhitworth/golearn.git synced 2025-04-25 13:48:49 +08:00
Code Issues Releases Wiki Activity

format

Browse Source
This commit is contained in:
frozenkp 2018-06-07 01:55:40 +08:00 committed by ss8651twtw
parent 1b9e563772
commit 6b3f7ddab2
20 changed files with 177 additions and 177 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

54
clustering/cluster_extra_test.go
View File

@ -71,42 +71,42 @@ func Test(t *testing.T) {
})
Convey("m[1] size diff", t, func() {
m1 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 3}
m1[0] = []int{4, 5}
m1 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 3}
m1[0] = []int{4, 5}
m2 := ClusterMap(make(map[int][]int))
m2[1] = []int{1, 2, 3}
m1[0] = []int{4, 5}
m2 := ClusterMap(make(map[int][]int))
m2[1] = []int{1, 2, 3}
m1[0] = []int{4, 5}
_, err := m1.Equals(m2)
So(err, ShouldNotBeNil)
})
_, err := m1.Equals(m2)
So(err, ShouldNotBeNil)
})
Convey("m[1] duplicate", t, func() {
m1 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 1}
m1[0] = []int{4, 5}
m1 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 1}
m1[0] = []int{4, 5}
m2 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 1}
m1[0] = []int{4, 5}
m2 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 1}
m1[0] = []int{4, 5}
_, err := m1.Equals(m2)
So(err, ShouldNotBeNil)
})
_, err := m1.Equals(m2)
So(err, ShouldNotBeNil)
})
Convey("m[0] duplicate", t, func() {
m1 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 2}
m1[0] = []int{4, 4}
m1 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 2}
m1[0] = []int{4, 4}
m2 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 2}
m1[0] = []int{4, 4}
m2 := ClusterMap(make(map[int][]int))
m1[1] = []int{1, 2}
m1[0] = []int{4, 4}
_, err := m1.Equals(m2)
So(err, ShouldNotBeNil)
})
_, err := m1.Equals(m2)
So(err, ShouldNotBeNil)
})
}

2
clustering/dbscan.go
View File

@ -1,9 +1,9 @@
package clustering
import (
"gonum.org/v1/gonum/mat"
"github.com/sjwhitworth/golearn/base"
"github.com/sjwhitworth/golearn/metrics/pairwise"
"gonum.org/v1/gonum/mat"
"math/big"
)

2
clustering/dbscan_test.go
View File

@ -2,10 +2,10 @@ package clustering
import (
"bufio"
"gonum.org/v1/gonum/mat"
"github.com/sjwhitworth/golearn/base"
"github.com/sjwhitworth/golearn/metrics/pairwise"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
"math"
"math/big"
"os"

36
clustering/em_test.go
View File

@ -10,25 +10,25 @@ func TestExpectationMaximization(t *testing.T) {
Convey("Doing EM-based clustering", t, func() {
em, _ := NewExpectationMaximization(2)
// Initialization tests
// Trying to create NewExpectationMaximization with < 1 component
Convey("With less than one component", func() {
Convey("Creating a new instance", func () {
_, err := NewExpectationMaximization(0)
// Initialization tests
// Trying to create NewExpectationMaximization with < 1 component
Convey("With less than one component", func() {
Convey("Creating a new instance", func() {
_, err := NewExpectationMaximization(0)
Convey("Should result in a InsufficientComponentsError", func() {
So(err, ShouldEqual, InsufficientComponentsError)
})
})
})
})
})
// Data tests
// Trying to Fit with fewer samples than components
// Data tests
// Trying to Fit with fewer samples than components
Convey("With insufficient training data", func() {
Convey("Fitting", func() {
testData, err := base.ParseCSVToInstances("./gaussian_mixture_single_obs.csv", false)
So(err, ShouldBeNil)
err = em.Fit(testData)
err = em.Fit(testData)
Convey("Should result in a InsufficientDataError", func() {
So(err, ShouldEqual, InsufficientDataError)
@ -36,7 +36,7 @@ func TestExpectationMaximization(t *testing.T) {
})
})
// Trying to Predict before having Fit
// Trying to Predict before having Fit
Convey("With no training data", func() {
Convey("Predicting", func() {
testData, err := base.ParseCSVToInstances("./gaussian_mixture.csv", false)
@ -50,8 +50,8 @@ func TestExpectationMaximization(t *testing.T) {
})
})
// Computation tests
// Test the predictions are resonable
// Computation tests
// Test the predictions are resonable
Convey("With sufficient training data", func() {
instances, err := base.ParseCSVToInstances("./gaussian_mixture.csv", true)
So(err, ShouldBeNil)
@ -60,10 +60,10 @@ func TestExpectationMaximization(t *testing.T) {
err := em.Fit(instances)
So(err, ShouldBeNil)
first_mean := em.Params.Means.At(0,0)
first_mean := em.Params.Means.At(0, 0)
Convey("It converges to reasonable a value", func() {
So(first_mean, ShouldAlmostEqual, -5.973, .1)
So(first_mean, ShouldAlmostEqual, -5.973, .1)
})
})
})
@ -76,9 +76,9 @@ func BenchmarkExpectationMaximizationOneRow(b *testing.B) {
testData, _ := base.ParseCSVToInstances("./gaussian_mixture.csv", false)
em, err := NewExpectationMaximization(2)
if err != nil {
panic(err)
}
if err != nil {
panic(err)
}
em.Fit(trainData)
b.ResetTimer()

22
kdtree/heap_test.go
View File

@ -8,25 +8,25 @@ import (
func TestHeap(t *testing.T) {
Convey("Given a heap", t, func() {
h := newHeap()
h := newHeap()
Convey("When heap is empty", func() {
Convey("The size should be 0", func() {
size := h.size()
size := h.size()
So(size, ShouldEqual, 0)
})
Convey("The maximum node should be empty", func(){
max := h.maximum()
So(max.value, ShouldBeEmpty)
So(max.length, ShouldEqual, 0)
So(max.srcRowNo, ShouldEqual, 0)
})
Convey("The maximum node should be empty", func() {
max := h.maximum()
So(max.value, ShouldBeEmpty)
So(max.length, ShouldEqual, 0)
So(max.srcRowNo, ShouldEqual, 0)
})
Convey("Extrace Max should be fine", func(){
So(func(){h.extractMax()}, ShouldNotPanic)
})
Convey("Extrace Max should be fine", func() {
So(func() { h.extractMax() }, ShouldNotPanic)
})
})
Convey("When insert 10 nodes", func() {

4
kdtree/kdtree.go
View File

@ -2,13 +2,13 @@ package kdtree
import (
"errors"
"gonum.org/v1/gonum/mat"
"github.com/sjwhitworth/golearn/metrics/pairwise"
"gonum.org/v1/gonum/mat"
"sort"
)
type node struct {
feature int // -1: leaf, -2: nil
feature int // -1: leaf, -2: nil
value []float64
srcRowNo int
left *node

206
kdtree/kdtree_test.go
View File

@ -8,123 +8,123 @@ import (
)
func TestKdtree(t *testing.T) {
Convey("Test Build", t, func(){
Convey("When no input data", func(){
kd := New()
data := [][]float64{}
err := kd.Build(data)
So(err.Error(), ShouldEqual, "no input data")
})
Convey("Test Build", t, func() {
Convey("When no input data", func() {
kd := New()
data := [][]float64{}
err := kd.Build(data)
So(err.Error(), ShouldEqual, "no input data")
})
Convey("When amounts of features not the same", func(){
kd := New()
data := [][]float64{{3, 5}, {6, 7, 10}}
err := kd.Build(data)
So(err.Error(), ShouldEqual, "amounts of features are not the same")
})
Convey("When amounts of features not the same", func() {
kd := New()
data := [][]float64{{3, 5}, {6, 7, 10}}
err := kd.Build(data)
So(err.Error(), ShouldEqual, "amounts of features are not the same")
})
Convey("When only one data", func(){
kd := New()
data := [][]float64{{3, 5}}
err := kd.Build(data)
So(err, ShouldBeNil)
})
Convey("When only one data", func() {
kd := New()
data := [][]float64{{3, 5}}
err := kd.Build(data)
So(err, ShouldBeNil)
})
Convey("When data all the same", func(){
kd := New()
data := [][]float64{{3, 5}, {3, 5}, {3, 5}}
err := kd.Build(data)
So(err, ShouldBeNil)
})
})
Convey("When data all the same", func() {
kd := New()
data := [][]float64{{3, 5}, {3, 5}, {3, 5}}
err := kd.Build(data)
So(err, ShouldBeNil)
})
})
Convey("Test Search", t, func() {
Convey("Functionally test", func(){
kd := New()
data := [][]float64{{2, 3}, {5, 4}, {4, 7}, {8, 1}, {7, 2}, {9, 6}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
Convey("Functionally test", func() {
kd := New()
data := [][]float64{{2, 3}, {5, 4}, {4, 7}, {8, 1}, {7, 2}, {9, 6}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
Convey("When k is 3 with euclidean", func() {
result, _, _ := kd.Search(3, euclidean, []float64{7, 3})
Convey("When k is 3 with euclidean", func() {
result, _, _ := kd.Search(3, euclidean, []float64{7, 3})
Convey("The result[0] should be 4", func() {
So(result[0], ShouldEqual, 4)
})
Convey("The result[1] should be 3", func() {
So(result[1], ShouldEqual, 3)
})
Convey("The result[2] should be 1", func() {
So(result[2], ShouldEqual, 1)
})
})
Convey("The result[0] should be 4", func() {
So(result[0], ShouldEqual, 4)
})
Convey("The result[1] should be 3", func() {
So(result[1], ShouldEqual, 3)
})
Convey("The result[2] should be 1", func() {
So(result[2], ShouldEqual, 1)
})
})
Convey("When k is 2 with euclidean", func() {
result, _, _ := kd.Search(2, euclidean, []float64{7, 3})
Convey("When k is 2 with euclidean", func() {
result, _, _ := kd.Search(2, euclidean, []float64{7, 3})
Convey("The result[0] should be 4", func() {
So(result[0], ShouldEqual, 4)
})
Convey("The result[1] should be 1", func() {
So(result[1], ShouldEqual, 1)
})
})
})
Convey("The result[0] should be 4", func() {
So(result[0], ShouldEqual, 4)
})
Convey("The result[1] should be 1", func() {
So(result[1], ShouldEqual, 1)
})
})
})
Convey("When k is larger than amount of trainData", func(){
kd := New()
data := [][]float64{{3, 5}, {2, 1}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
_, _, err := kd.Search(3, euclidean, []float64{7, 3})
So(err.Error(), ShouldEqual, "k is largerer than amount of trainData")
})
Convey("When k is larger than amount of trainData", func() {
kd := New()
data := [][]float64{{3, 5}, {2, 1}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
_, _, err := kd.Search(3, euclidean, []float64{7, 3})
So(err.Error(), ShouldEqual, "k is largerer than amount of trainData")
})
Convey("When features of target is larger than trainData", func(){
kd := New()
data := [][]float64{{3, 5}, {2, 1}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
_, _, err := kd.Search(1, euclidean, []float64{7, 3, 5})
So(err.Error(), ShouldEqual, "amount of features is not equal")
})
Convey("When features of target is larger than trainData", func() {
kd := New()
data := [][]float64{{3, 5}, {2, 1}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
_, _, err := kd.Search(1, euclidean, []float64{7, 3, 5})
So(err.Error(), ShouldEqual, "amount of features is not equal")
})
Convey("When node.feature is -2", func(){
kd := New()
data := [][]float64{{3, 5}, {2, 1}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
_, _, err := kd.Search(1, euclidean, []float64{7, 3})
So(err, ShouldBeNil)
})
Convey("When node.feature is -2", func() {
kd := New()
data := [][]float64{{3, 5}, {2, 1}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
_, _, err := kd.Search(1, euclidean, []float64{7, 3})
So(err, ShouldBeNil)
})
Convey("Search All Node (left)", func(){
kd := New()
data := [][]float64{{1, 2}, {5, 6}, {9, 10}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
result, _, _ := kd.Search(1, euclidean, []float64{7, 3})
So(result[0], ShouldEqual, 1)
})
Convey("Search All Node (left)", func() {
kd := New()
data := [][]float64{{1, 2}, {5, 6}, {9, 10}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
result, _, _ := kd.Search(1, euclidean, []float64{7, 3})
So(result[0], ShouldEqual, 1)
})
Convey("Search when node length larger than heap max", func(){
Convey("Search All Node (left)", func(){
kd := New()
data := [][]float64{{1, 2}, {5, 6}, {9, 10}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
result, _, _ := kd.Search(1, euclidean, []float64{8, 7})
So(result[0], ShouldEqual, 2)
})
Convey("Search when node length larger than heap max", func() {
Convey("Search All Node (left)", func() {
kd := New()
data := [][]float64{{1, 2}, {5, 6}, {9, 10}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
result, _, _ := kd.Search(1, euclidean, []float64{8, 7})
So(result[0], ShouldEqual, 2)
})
Convey("Search All Node (right)", func(){
kd := New()
data := [][]float64{{1, 2}, {5, 4}, {9, 10}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
result, _, _ := kd.Search(1, euclidean, []float64{3, 3})
So(result[0], ShouldEqual, 0)
})
})
Convey("Search All Node (right)", func() {
kd := New()
data := [][]float64{{1, 2}, {5, 4}, {9, 10}}
kd.Build(data)
euclidean := pairwise.NewEuclidean()
result, _, _ := kd.Search(1, euclidean, []float64{3, 3})
So(result[0], ShouldEqual, 0)
})
})
})
}

2
knn/knn.go
View File

@ -8,11 +8,11 @@ import (
"fmt"
"github.com/gonum/matrix"
"gonum.org/v1/gonum/mat"
"github.com/sjwhitworth/golearn/base"
"github.com/sjwhitworth/golearn/kdtree"
"github.com/sjwhitworth/golearn/metrics/pairwise"
"github.com/sjwhitworth/golearn/utilities"
"gonum.org/v1/gonum/mat"
)
// A KNNClassifier consists of a data matrix, associated labels in the same order as the matrix, searching algorithm, and a distance function.

4
meta/bagging.go
View File

@ -253,7 +253,7 @@ func (b *BaggedModel) SaveWithPrefix(writer *base.ClassifierSerializer, prefix s
// Save the classifiers
for i, c := range b.Models {
clsPrefix := fmt.Sprintf("%s/", pI( "CLASSIFIERS", i))
clsPrefix := fmt.Sprintf("%s/", pI("CLASSIFIERS", i))
err = c.SaveWithPrefix(writer, clsPrefix)
if err != nil {
return base.FormatError(err, "Can't save classifier %d", i)
@ -308,7 +308,7 @@ func (b *BaggedModel) LoadWithPrefix(reader *base.ClassifierDeserializer, prefix
// Reload the classifiers
for i, m := range b.Models {
clsPrefix := fmt.Sprintf("%s/", pI( "CLASSIFIERS", i))
clsPrefix := fmt.Sprintf("%s/", pI("CLASSIFIERS", i))
err := m.LoadWithPrefix(reader, clsPrefix)
if err != nil {
return base.DescribeError("Can't read classifier", err)

2
metrics/pairwise/chebyshev_test.go
View File

@ -3,8 +3,8 @@ package pairwise
import (
"testing"
"gonum.org/v1/gonum/mat"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
)
func TestChebyshev(t *testing.T) {

2
metrics/pairwise/cosine_test.go
View File

@ -3,8 +3,8 @@ package pairwise
import (
"testing"
"gonum.org/v1/gonum/mat"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
)
func TestCosine(t *testing.T) {

2
metrics/pairwise/cranberra_test.go
View File

@ -3,8 +3,8 @@ package pairwise
import (
"testing"
"gonum.org/v1/gonum/mat"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
)
func TestCranberrra(t *testing.T) {

2
metrics/pairwise/euclidean_test.go
View File

@ -3,8 +3,8 @@ package pairwise
import (
"testing"
"gonum.org/v1/gonum/mat"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
)
func TestEuclidean(t *testing.T) {

2
metrics/pairwise/manhattan_test.go
View File

@ -3,8 +3,8 @@ package pairwise
import (
"testing"
"gonum.org/v1/gonum/mat"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
)
func TestManhattan(t *testing.T) {

2
metrics/pairwise/poly_kernel_test.go
View File

@ -3,8 +3,8 @@ package pairwise
import (
"testing"
"gonum.org/v1/gonum/mat"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
)
func TestPolyKernel(t *testing.T) {

2
metrics/pairwise/rbf_kernel_test.go
View File

@ -3,8 +3,8 @@ package pairwise
import (
"testing"
"gonum.org/v1/gonum/mat"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
)
func TestRBFKernel(t *testing.T) {

2
neural/layered.go
View File

@ -2,9 +2,9 @@ package neural
import (
"fmt"
"gonum.org/v1/gonum/mat"
"github.com/sjwhitworth/golearn/base"
"github.com/sjwhitworth/golearn/filters"
"gonum.org/v1/gonum/mat"
"math"
"math/rand"
)

2
neural/layered_test.go
View File

@ -1,9 +1,9 @@
package neural
import (
"gonum.org/v1/gonum/mat"
"github.com/sjwhitworth/golearn/base"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
"testing"
)

2
neural/network.go
View File

@ -12,7 +12,7 @@ import (
// NeuralFunction.
type Network struct {
origWeights *mat.Dense
weights *mat.Dense // n * n
weights *mat.Dense // n * n
biases []float64 // n for each neuron
funcs []NeuralFunction // for each neuron
size int

2
neural/network_test.go
View File

@ -1,8 +1,8 @@
package neural
import (
"gonum.org/v1/gonum/mat"
. "github.com/smartystreets/goconvey/convey"
"gonum.org/v1/gonum/mat"
"testing"
)