Fix Issue #5 - Variable Logging to io.Writers

Switched all models that did print some
sort of logging to using any io.Writer.
This gives the user control over whether
to log to stdout (which is default), log
to some file or API or something, or not
log at all. Every model that did print
has a public struct field such as below
(only the comment might differ slightly. The
name is always `Output io.Writer`

```go
        // Output is the io.Writer used for logging
        // and printing. Defaults to os.Stdout.
        Output io.Writer
```

Author: cdipaolo

base/data.go

@@ -49,7 +49,6 @@ func LoadDataFromCSV(filepath string) ([][]float64, []float64, error) {
 		return nil, nil, err
 	}
 
-	fmt.Printf("Loading Data From CSV <%v>\n", filepath)
 	// parse until the end of the file
 	for err != io.EOF {
 		var row []float64
@@ -76,7 +75,6 @@ func LoadDataFromCSV(filepath string) ([][]float64, []float64, error) {
 		return nil, nil, fmt.Errorf("ERROR: Training set has no valid examples (either for x or y or both)")
 	}
 
-	fmt.Printf("Finished Loading Data From <%v>\n\tTraining Examples: %v\n\tFeatures: %v\n", filepath, len(y), len(x[0]))
 	return x, y, nil
 }
 
@@ -122,7 +120,6 @@ func LoadDataFromCSVToStream(filepath string, data chan Datapoint, errors chan e
 		return
 	}
 
-	fmt.Printf("Loading Data From CSV <%v> Into Data Channel\n", filepath)
 	// parse until the end of the file
 	for err != io.EOF {
 		var row []float64
@@ -149,8 +146,6 @@ func LoadDataFromCSVToStream(filepath string, data chan Datapoint, errors chan e
 		record, err = reader.Read()
 	}
 
-	fmt.Printf("Finished Loading Data From <%v> Into Data Channel\n\tClosing error channel\n\tClosing data channel\n", filepath)
-
 	close(errors)
 	close(data)
 	return
@@ -201,7 +196,6 @@ func SaveDataToCSV(filepath string, x [][]float64, y []float64, highPrecision bo
 	writer := csv.NewWriter(file)
 	records := [][]string{}
 
-	fmt.Printf("Writing Data To <%v>\n\tTraining Examples: %v\n\tFeatures: %v\n", filepath, len(x), len(x[0]))
 	// parse until the end of the file
 	for i := range x {
 		record := []string{}
@@ -221,6 +215,5 @@ func SaveDataToCSV(filepath string, x [][]float64, y []float64, highPrecision bo
 		return err
 	}
 
-	fmt.Printf("Finished Writing Data To <%v>\n\n", filepath)
 	return nil
 }

base/munge.go

@@ -1,7 +1,6 @@
 package base
 
 import (
-	"fmt"
 	"math"
 )
 
@@ -16,8 +15,6 @@ func Normalize(x [][]float64) {
 	for i := range x {
 		NormalizePoint(x[i])
 	}
-
-	fmt.Printf("Normalized < %v > data points by dividing by unit length\n", len(x))
 }
 
 // NormalizePoint is the same as Normalize,

base/optimize.go

@@ -52,8 +52,6 @@ func GradientAscent(d Ascendable) error {
 		}
 	}
 
-	fmt.Printf("Went through %v iterations.\n", iter+1)
-
 	return nil
 }
 
@@ -114,7 +112,5 @@ func StochasticGradientAscent(d StochasticAscendable) error {
 		}
 	}
 
-	fmt.Printf("Went through %v iterations.\n", iter+1)
-
 	return nil
 }

cluster/kmeans.go

@@ -3,6 +3,7 @@ package cluster
 import (
 	"encoding/json"
 	"fmt"
+	"io"
 	"io/ioutil"
 	"math/rand"
 	"os"
@@ -133,6 +134,11 @@ type KMeans struct {
 	guesses     []int
 
 	Centroids [][]float64 `json:"centroids"`
+
+	// Output is the io.Writer to write
+	// logging to. Defaults to os.Stdout
+	// but can be changed to any io.Writer
+	Output io.Writer
 }
 
 // OnlineParams is used to pass optional
@@ -190,6 +196,7 @@ func NewKMeans(k, maxIterations int, trainingSet [][]float64, params ...OnlinePa
 		guesses:     guesses,
 
 		Centroids: centroids,
+		Output:    os.Stdout,
 	}
 }
 
@@ -275,21 +282,21 @@ func (k *KMeans) Predict(x []float64, normalize ...bool) ([]float64, error) {
 func (k *KMeans) Learn() error {
 	if k.trainingSet == nil {
 		err := fmt.Errorf("ERROR: Attempting to learn with no training examples!\n")
-		fmt.Printf(err.Error())
+		fmt.Fprintf(k.Output, err.Error())
 		return err
 	}
 
 	examples := len(k.trainingSet)
 	if examples == 0 || len(k.trainingSet[0]) == 0 {
 		err := fmt.Errorf("ERROR: Attempting to learn with no training examples!\n")
-		fmt.Printf(err.Error())
+		fmt.Fprintf(k.Output, err.Error())
 		return err
 	}
 
 	centroids := len(k.Centroids)
 	features := len(k.trainingSet[0])
 
-	fmt.Printf("Training:\n\tModel: K-Means++ Classification\n\tTraining Examples: %v\n\tFeatures: %v\n\tClasses: %v\n...\n\n", examples, features, centroids)
+	fmt.Fprintf(k.Output, "Training:\n\tModel: K-Means++ Classification\n\tTraining Examples: %v\n\tFeatures: %v\n\tClasses: %v\n...\n\n", examples, features, centroids)
 
 	// instantiate the centroids using k-means++
 	k.Centroids[0] = k.trainingSet[rand.Intn(len(k.trainingSet))]
@@ -372,7 +379,7 @@ func (k *KMeans) Learn() error {
 		}
 	}
 
-	fmt.Printf("Training Completed in %v iterations.\n%v\n", iter, k)
+	fmt.Fprintf(k.Output, "Training Completed in %v iterations.\n%v\n", iter, k)
 
 	return nil
 }
@@ -499,7 +506,7 @@ func (k *KMeans) OnlineLearn(errors chan error, dataset chan base.Datapoint, onU
 	centroids := len(k.Centroids)
 	features := len(k.Centroids[0])
 
-	fmt.Printf("Training:\n\tModel: Online K-Means Classification\n\tFeatures: %v\n\tClasses: %v\n...\n\n", features, centroids)
+	fmt.Fprintf(k.Output, "Training:\n\tModel: Online K-Means Classification\n\tFeatures: %v\n\tClasses: %v\n...\n\n", features, centroids)
 
 	var point base.Datapoint
 	var more bool
@@ -531,7 +538,7 @@ func (k *KMeans) OnlineLearn(errors chan error, dataset chan base.Datapoint, onU
 			go onUpdate([][]float64{[]float64{float64(c)}, k.Centroids[c]})
 
 		} else {
-			fmt.Printf("Training Completed.\n%v\n\n", k)
+			fmt.Fprintf(k.Output, "Training Completed.\n%v\n\n", k)
 			close(errors)
 			return
 		}

cluster/triangle_kmeans.go

@@ -3,6 +3,7 @@ package cluster
 import (
 	"encoding/json"
 	"fmt"
+	"io"
 	"io/ioutil"
 	"math/rand"
 	"os"
@@ -146,6 +147,10 @@ type TriangleKMeans struct {
 	// calculations
 	centroidDist    [][]float64
 	minCentroidDist []float64
+
+	// Output is the io.Writer to write logs
+	// and output from training to
+	Output io.Writer
 }
 
 // pointInfo stores information needed to use
@@ -214,6 +219,8 @@ func NewTriangleKMeans(k, maxIterations int, trainingSet [][]float64) *TriangleK
 		Centroids:       centroids,
 		centroidDist:    centroidDist,
 		minCentroidDist: minCentroidDist,
+
+		Output: os.Stdout,
 	}
 }
 
@@ -383,21 +390,21 @@ func (k *TriangleKMeans) recalculateCentroids() [][]float64 {
 func (k *TriangleKMeans) Learn() error {
 	if k.trainingSet == nil {
 		err := fmt.Errorf("ERROR: Attempting to learn with no training examples!\n")
-		fmt.Printf(err.Error())
+		fmt.Fprintf(k.Output, err.Error())
 		return err
 	}
 
 	examples := len(k.trainingSet)
 	if examples == 0 || len(k.trainingSet[0]) == 0 {
 		err := fmt.Errorf("ERROR: Attempting to learn with no training examples!\n")
-		fmt.Printf(err.Error())
+		fmt.Fprintf(k.Output, err.Error())
 		return err
 	}
 
 	centroids := len(k.Centroids)
 	features := len(k.trainingSet[0])
 
-	fmt.Printf("Training:\n\tModel: Triangle Inequality Accelerated K-Means++ Classification\n\tTraining Examples: %v\n\tFeatures: %v\n\tClasses: %v\n...\n\n", examples, features, centroids)
+	fmt.Fprintf(k.Output, "Training:\n\tModel: Triangle Inequality Accelerated K-Means++ Classification\n\tTraining Examples: %v\n\tFeatures: %v\n\tClasses: %v\n...\n\n", examples, features, centroids)
 
 	/* Step 0 */
 
@@ -535,7 +542,7 @@ func (k *TriangleKMeans) Learn() error {
 		k.Centroids = newCentroids
 	}
 
-	fmt.Printf("Training Completed in %v iterations.\n%v\n", iter, k)
+	fmt.Fprintf(k.Output, "Training Completed in %v iterations.\n%v\n", iter, k)
 
 	return nil
 }

linear/linear.go

@@ -53,6 +53,7 @@ import (
 	"bytes"
 	"encoding/json"
 	"fmt"
+	"io"
 	"io/ioutil"
 	"math"
 	"os"
@@ -92,6 +93,10 @@ type LeastSquares struct {
 	expectedResults []float64
 
 	Parameters []float64 `json:"theta"`
+
+	// Output is the io.Writer used for logging
+	// and printing. Defaults to os.Stdout.
+	Output io.Writer
 }
 
 // NewLeastSquares returns a pointer to the linear model
@@ -148,6 +153,8 @@ func NewLeastSquares(method base.OptimizationMethod, alpha, regularization float
 		// initialize θ as the zero vector (that is,
 		// the vector of all zeros)
 		Parameters: params,
+
+		Output: os.Stdout,
 	}
 }
 
@@ -229,23 +236,23 @@ func (l *LeastSquares) Predict(x []float64, normalize ...bool) ([]float64, error
 func (l *LeastSquares) Learn() error {
 	if l.trainingSet == nil || l.expectedResults == nil {
 		err := fmt.Errorf("ERROR: Attempting to learn with no training examples!\n")
-		fmt.Printf(err.Error())
+		fmt.Fprintf(l.Output, err.Error())
 		return err
 	}
 
 	examples := len(l.trainingSet)
 	if examples == 0 || len(l.trainingSet[0]) == 0 {
 		err := fmt.Errorf("ERROR: Attempting to learn with no training examples!\n")
-		fmt.Printf(err.Error())
+		fmt.Fprintf(l.Output, err.Error())
 		return err
 	}
 	if len(l.expectedResults) == 0 {
 		err := fmt.Errorf("ERROR: Attempting to learn with no expected results! This isn't an unsupervised model!! You'll need to include data before you learn :)\n")
-		fmt.Printf(err.Error())
+		fmt.Fprintf(l.Output, err.Error())
 		return err
 	}
 
-	fmt.Printf("Training:\n\tModel: Logistic (Binary) Classification\n\tOptimization Method: %v\n\tTraining Examples: %v\n\tFeatures: %v\n\tLearning Rate α: %v\n\tRegularization Parameter λ: %v\n...\n\n", l.method, examples, len(l.trainingSet[0]), l.alpha, l.regularization)
+	fmt.Fprintf(l.Output, "Training:\n\tModel: Logistic (Binary) Classification\n\tOptimization Method: %v\n\tTraining Examples: %v\n\tFeatures: %v\n\tLearning Rate α: %v\n\tRegularization Parameter λ: %v\n...\n\n", l.method, examples, len(l.trainingSet[0]), l.alpha, l.regularization)
 
 	var err error
 	if l.method == base.BatchGA {
@@ -257,11 +264,11 @@ func (l *LeastSquares) Learn() error {
 	}
 
 	if err != nil {
-		fmt.Printf("\nERROR: Error while learning –\n\t%v\n\n", err)
+		fmt.Fprintf(l.Output, "\nERROR: Error while learning –\n\t%v\n\n", err)
 		return err
 	}
 
-	fmt.Printf("Training Completed.\n%v\n\n", l)
+	fmt.Fprintf(l.Output, "Training Completed.\n%v\n\n", l)
 	return nil
 }
 
@@ -368,7 +375,7 @@ func (l *LeastSquares) OnlineLearn(errors chan error, dataset chan base.Datapoin
 		return
 	}
 
-	fmt.Printf("Training:\n\tModel: Ordinary Least Squares Regression\n\tOptimization Method: Online Stochastic Gradient Descent\n\tFeatures: %v\n\tLearning Rate α: %v\n...\n\n", len(l.Parameters), l.alpha)
+	fmt.Fprintf(l.Output, "Training:\n\tModel: Ordinary Least Squares Regression\n\tOptimization Method: Online Stochastic Gradient Descent\n\tFeatures: %v\n\tLearning Rate α: %v\n...\n\n", len(l.Parameters), l.alpha)
 
 	var point base.Datapoint
 	var more bool
@@ -439,7 +446,7 @@ func (l *LeastSquares) OnlineLearn(errors chan error, dataset chan base.Datapoin
 			go onUpdate([][]float64{l.Parameters})
 
 		} else {
-			fmt.Printf("Training Completed.\n%v\n\n", l)
+			fmt.Fprintf(l.Output, "Training Completed.\n%v\n\n", l)
 			close(errors)
 			return
 		}
@@ -452,7 +459,7 @@ func (l *LeastSquares) OnlineLearn(errors chan error, dataset chan base.Datapoin
 func (l *LeastSquares) String() string {
 	features := len(l.Parameters) - 1
 	if len(l.Parameters) == 0 {
-		fmt.Printf("ERROR: Attempting to print model with the 0 vector as it's parameter vector! Train first!\n")
+		fmt.Fprintf(l.Output, "ERROR: Attempting to print model with the 0 vector as it's parameter vector! Train first!\n")
 	}
 	var buffer bytes.Buffer
 

linear/local_linear.go

@@ -3,7 +3,9 @@ package linear
 import (
 	"bytes"
 	"fmt"
+	"io"
 	"math"
+	"os"
 
 	"github.com/cdipaolo/goml/base"
 )
@@ -100,6 +102,10 @@ type LocalLinear struct {
 	expectedResults []float64
 
 	Parameters []float64 `json:"theta"`
+
+	// Output is the io.Writer used for logging
+	// and printing. Defaults to os.Stdout.
+	Output io.Writer
 }
 
 // NewLocalLinear returns a pointer to the linear model
@@ -156,6 +162,8 @@ func NewLocalLinear(method base.OptimizationMethod, alpha, regularization, bandw
 		// initialize θ as the zero vector (that is,
 		// the vector of all zeros)
 		Parameters: params,
+
+		Output: os.Stdout,
 	}
 }
 
@@ -242,7 +250,7 @@ func (l *LocalLinear) Predict(x []float64, normalize ...bool) ([]float64, error)
 		return nil, err
 	}
 
-	fmt.Printf("Training:\n\tModel: Locally Weighted Linear Regression\n\tOptimization Method: %v\n\tCenter Point: %v\n\tTraining Examples: %v\n\tFeatures: %v\n\tLearning Rate α: %v\n\tRegularization Parameter λ: %v\n...\n\n", l.method, x, examples, len(l.trainingSet[0]), l.alpha, l.regularization)
+	fmt.Fprintf(l.Output, "Training:\n\tModel: Locally Weighted Linear Regression\n\tOptimization Method: %v\n\tCenter Point: %v\n\tTraining Examples: %v\n\tFeatures: %v\n\tLearning Rate α: %v\n\tRegularization Parameter λ: %v\n...\n\n", l.method, x, examples, len(l.trainingSet[0]), l.alpha, l.regularization)
 
 	var iter int
 	features := len(l.Parameters)
@@ -295,7 +303,7 @@ func (l *LocalLinear) Predict(x []float64, normalize ...bool) ([]float64, error)
 		return nil, fmt.Errorf("Chose a training method not implemented for LocalLinear regression")
 	}
 
-	fmt.Printf("Training Completed. Went through %v iterations.\n%v\n\n", iter, l)
+	fmt.Fprintf(l.Output, "Training Completed. Went through %v iterations.\n%v\n\n", iter, l)
 
 	// include constant term in sum
 	sum := l.Parameters[0]
@@ -313,7 +321,7 @@ func (l *LocalLinear) Predict(x []float64, normalize ...bool) ([]float64, error)
 func (l *LocalLinear) String() string {
 	features := len(l.Parameters) - 1
 	if len(l.Parameters) == 0 {
-		fmt.Printf("ERROR: Attempting to print model with the 0 vector as it's parameter vector! Train first!\n")
+		fmt.Fprintf(l.Output, "ERROR: Attempting to print model with the 0 vector as it's parameter vector! Train first!\n")
 	}
 	var buffer bytes.Buffer