Starting interface design

base/predictor.go

@@ -0,0 +1,15 @@
+package base
+
+// Specifies the base interfaces
+type Estimator interface {
+	Fit()
+	Summarise()
+}
+
+type Predictor interface {
+	Predict()
+}
+
+type Model interface {
+	Score()
+}

data/csv.go

@@ -0,0 +1,57 @@
+/* Data - consists of helper functions for parsing different data formats */
+
+package data
+
+import (
+	"encoding/csv"
+	"fmt"
+	"io"
+	"os"
+	"strconv"
+)
+
+//Parses a CSV file, returning the number of columns and rows, the headers, the labels associated with
+//classification, and the data that will be used for training.
+func ParseCsv(filepath string, label int, columns []int) (int, int, []string, []string, []float64) {
+	labels := make([]string, 0)
+	data := make([]float64, 0)
+	headers := make([]string, 0)
+	rows := 0
+
+	file, err := os.Open(filepath)
+	if err != nil {
+		fmt.Println("Error:", err)
+	}
+	defer file.Close()
+
+	reader := csv.NewReader(file)
+
+	headerrow, _ := reader.Read()
+
+	for _, col := range columns {
+		entry := headerrow[col]
+		headers = append(headers, entry)
+	}
+
+	for {
+		record, err := reader.Read()
+		if err == io.EOF {
+			break
+		} else if err != nil {
+			fmt.Println("Error:", err)
+		}
+
+		//
+		labels = append(labels, record[label])
+
+		//Iterate over our rows and append the values to a slice
+		for _, col := range columns {
+			entry := record[col]
+			number, _ := strconv.ParseFloat(entry, 64)
+			data = append(data, number)
+		}
+		rows += 1
+	}
+	cols := len(columns)
+	return cols, rows, headers, labels, data
+}

examples/knnclassifier_iris.go

@@ -1,31 +1,30 @@
 package main
 
 import (
-		mat "github.com/skelterjohn/go.matrix"
-		base "golearn/base"
-		util "golearn/utilities"
-		knnclass "golearn/knn"
-		"fmt"
-		)
+	"fmt"
+	mat "github.com/skelterjohn/go.matrix"
+	"golearn/data"
+	util "golearn/utilities"
+)
 
-func main(){
+func main() {
 	//Parses the infamous Iris data.
-	cols, rows, _, labels, data := base.ParseCsv("datasets/iris.csv", 4, []int{0,1,2})
+	cols, rows, _, labels, data := base.ParseCsv("datasets/iris.csv", 4, []int{0, 1, 2})
 
 	//Initialises a new KNN classifier
 	knn := knnclass.KNNClassifier{}
 	knn.C
 	knn.New("Testing", labels, data, rows, cols)
-	
+
 	for {
 		//Creates a random array of N float64s between 0 and 7
 		randArray := util.RandomArray(3, 7)
 
 		//Initialises a vector with this array
-		random := mat.MakeDenseMatrix(randArray,1,3)
+		random := mat.MakeDenseMatrix(randArray, 1, 3)
 
 		//Calculates the Euclidean distance and returns the most popular label
 		labels, _ := knn.Predict(random, 3)
 		fmt.Println(labels)
 	}
-}
+}

knn/knn.go

@@ -1,12 +1,12 @@
 package knn
 
 import (
-		mat "github.com/skelterjohn/go.matrix"
-		"math"
-		"fmt"
-		util "golearn/utilities"
-		base "golearn/base"
-		)
+	"fmt"
+	mat "github.com/skelterjohn/go.matrix"
+	base "golearn/base"
+	util "golearn/utilities"
+	"math"
+)
 
 //A KNN Classifier. Consists of a data matrix, associated labels in the same order as the matrix, and a name.
 type KNNClassifier struct {
@@ -15,11 +15,6 @@ type KNNClassifier struct {
 
 //Mints a new classifier.
 func (KNN *KNNClassifier) New(name string, labels []string, numbers []float64, x int, y int) {
-	
-	//Write in some error handling here
-	// if x != len(KNN.Labels) {
-	// 	return errors.New("KNN: There must be a label for each row")
-	// }
 
 	KNN.Data = *mat.MakeDenseMatrix(numbers, x, y)
 	KNN.Name = name
@@ -54,7 +49,7 @@ func (KNN *KNNClassifier) Predict(vector *mat.DenseMatrix, K int) (string, []int
 	labels := make([]string, 0)
 	maxmap := make(map[string]int)
 
-	for i := 0; i < rows; i++{
+	for i := 0; i < rows; i++ {
 		row := KNN.Data.GetRowVector(i)
 		eucdistance := KNN.ComputeDistance(row, vector)
 		rownumbers[i] = eucdistance
@@ -77,4 +72,4 @@ func (KNN *KNNClassifier) Predict(vector *mat.DenseMatrix, K int) (string, []int
 	label := sortedlabels[0]
 
 	return label, values
-}
+}

knn/knnregressor.go

@@ -1,12 +1,12 @@
 package knn
 
 import (
-		mat "github.com/skelterjohn/go.matrix"
-		"math"
-		"fmt"
-		util "golearn/utilities"
-		base "golearn/base"
-		)
+	"fmt"
+	mat "github.com/skelterjohn/go.matrix"
+	base "golearn/base"
+	util "golearn/utilities"
+	"math"
+)
 
 //A KNN Regressor. Consists of a data matrix, associated result variables in the same order as the matrix, and a name.
 type KNNRegressor struct {
@@ -49,7 +49,7 @@ func (KNN *KNNRegressor) Predict(vector *mat.DenseMatrix, K int) (float64, []int
 	labels := make([]float64, 1)
 	sum := 0.0
 
-	for i := 0; i < rows; i++{
+	for i := 0; i < rows; i++ {
 		row := KNN.Data.GetRowVector(i)
 		eucdistance := KNN.ComputeDistance(row, vector)
 		rownumbers[i] = eucdistance
@@ -66,4 +66,4 @@ func (KNN *KNNRegressor) Predict(vector *mat.DenseMatrix, K int) (float64, []int
 
 	average := sum / float64(K)
 	return average, values
-}
+}

lm/linear_regression.go

@@ -0,0 +1,13 @@
+package lm
+
+import (
+	"fmt"
+	mat "github.com/skelterjohn/go.matrix"
+	base "golearn/base"
+	util "golearn/utilities"
+	"math"
+)
+
+type LinearModel struct {
+	base.BaseRegressor
+}

utilities/utilities.go

@@ -1,29 +1,29 @@
 package utilities
 
 import (
-		"sort"
-		rand "math/rand"
-		"fmt"
-		"strconv"
-		)
+	"fmt"
+	rand "math/rand"
+	"sort"
+	"strconv"
+)
 
 type sortedIntMap struct {
 	m map[int]float64
 	s []int
 }
- 
+
 func (sm *sortedIntMap) Len() int {
 	return len(sm.m)
 }
- 
+
 func (sm *sortedIntMap) Less(i, j int) bool {
 	return sm.m[sm.s[i]] < sm.m[sm.s[j]]
 }
- 
+
 func (sm *sortedIntMap) Swap(i, j int) {
 	sm.s[i], sm.s[j] = sm.s[j], sm.s[i]
 }
- 
+
 func SortIntMap(m map[int]float64) []int {
 	sm := new(sortedIntMap)
 	sm.m = m
@@ -41,19 +41,19 @@ type sortedStringMap struct {
 	m map[string]int
 	s []string
 }
- 
+
 func (sm *sortedStringMap) Len() int {
 	return len(sm.m)
 }
- 
+
 func (sm *sortedStringMap) Less(i, j int) bool {
 	return sm.m[sm.s[i]] < sm.m[sm.s[j]]
 }
- 
+
 func (sm *sortedStringMap) Swap(i, j int) {
 	sm.s[i], sm.s[j] = sm.s[j], sm.s[i]
 }
- 
+
 func SortStringMap(m map[string]int) []string {
 	sm := new(sortedStringMap)
 	sm.m = m
@@ -87,4 +87,4 @@ func ConvertLabelsToFloat(labels []string) []float64 {
 		floats = append(floats, converted)
 	}
 	return floats
-}
+}