classifier

package
v0.54.0 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Apr 4, 2024 License: BSD-3-Clause, BSD-3-Clause Imports: 10 Imported by: 0

Documentation

Overview

Package classifier provides machine learning classifier library, including CART, Random Forest, Cascaded Random Forest, and KNN.

Index

Constants

This section is empty.

Variables

This section is empty.

Functions

func ComputeAccuracies 

func ComputeAccuracies(tp, fp, tn, fn []int64) (accuracies []float64)

ComputeAccuracies will compute and return accuracy from array of true-positive, false-positive, true-negative, and false-negative; using formula, 

(tp + tn) / (tp + tn + tn + fn)

func ComputeElapsedTimes 

func ComputeElapsedTimes(start, end []int64) (elaps []int64)

ComputeElapsedTimes will compute and return elapsed time between `start` and `end` timestamps.

func ComputeFMeasures 

func ComputeFMeasures(precisions, recalls []float64) (fmeasures []float64)

ComputeFMeasures given array of precisions and recalls, compute F-measure of each instance and return it.

Types

type CM 

type CM struct {
	tabula.Dataset
	// contains filtered or unexported fields
}

CM represent the matrix of classification.

func (*CM) ComputeNumeric 

func (cm *CM) ComputeNumeric(vs, actuals, predictions []int64)

ComputeNumeric will calculate confusion matrix using targets and predictions values.

func (*CM) ComputeStrings 

func (cm *CM) ComputeStrings(valueSpace, targets, predictions []string)

ComputeStrings will calculate confusion matrix using targets and predictions class values.

func (*CM) FN 

func (cm *CM) FN() int

FN return number of false-negative.

func (*CM) FNIndices 

func (cm *CM) FNIndices() []int

FNIndices return indices of all false-negative samples.

func (*CM) FP 

func (cm *CM) FP() int

FP return number of false-positive in confusion matrix.

func (*CM) FPIndices 

func (cm *CM) FPIndices() []int

FPIndices return indices of all false-positive samples.

func (*CM) GetColumnClassError 

func (cm *CM) GetColumnClassError() *tabula.Column

GetColumnClassError return the last column which is the column that contain the error of classification.

func (*CM) GetFalseRate 

func (cm *CM) GetFalseRate() float64

GetFalseRate return false-positive rate in term of, 

false-positive / (false-positive + true negative)

func (*CM) GetTrueRate 

func (cm *CM) GetTrueRate() float64

GetTrueRate return true-positive rate in term of 

true-positive / (true-positive + false-positive)

func (*CM) GroupIndexPredictions 

func (cm *CM) GroupIndexPredictions(sampleListID []int,
	actuals, predictions []int64,
)

GroupIndexPredictions given index of samples, group the samples by their class of prediction. For example, 

sampleListID:   [0, 1, 2, 3, 4, 5]
actuals:     [1, 1, 0, 0, 1, 0]
predictions: [1, 0, 1, 0, 1, 1]

This function will group the index by true-positive, false-positive, true-negative, and false-negative, which result in, 

	true-positive indices:  [0, 4]
	false-positive indices: [2, 5]
	true-negative indices:  [3]
     false-negative indices: [1]

This function assume that positive value as "1" and negative value as "0".

func (*CM) GroupIndexPredictionsStrings 

func (cm *CM) GroupIndexPredictionsStrings(sampleListID []int,
	actuals, predictions []string,
)

GroupIndexPredictionsStrings is an alternative to GroupIndexPredictions which work with string class.

func (*CM) String 

func (cm *CM) String() (s string)

String will return the output of confusion matrix in table like format.

func (*CM) TN 

func (cm *CM) TN() int

TN return number of true-negative.

func (*CM) TNIndices 

func (cm *CM) TNIndices() []int

TNIndices return indices of all true-negative samples.

func (*CM) TP 

func (cm *CM) TP() int

TP return number of true-positive in confusion matrix.

func (*CM) TPIndices 

func (cm *CM) TPIndices() []int

TPIndices return indices of all true-positive samples.

type Runtime 

type Runtime struct {

	// OOBStatsFile is the file where OOB statistic will be written.
	OOBStatsFile string `json:"OOBStatsFile"`

	// PerfFile is the file where statistic of performance will be written.
	PerfFile string `json:"PerfFile"`

	// StatFile is the file where statistic of classifying samples will be
	// written.
	StatFile string `json:"StatFile"`

	// RunOOB if its true the OOB will be computed, default is false.
	RunOOB bool `json:"RunOOB"`
	// contains filtered or unexported fields
}

Runtime define a generic type which provide common fields that can be embedded by the real classifier (e.g. RandomForest).

func (*Runtime) AddOOBCM 

func (rt *Runtime) AddOOBCM(cm *CM)

AddOOBCM will append new confusion matrix.

func (*Runtime) AddStat 

func (rt *Runtime) AddStat(stat *Stat)

AddStat will append new classifier statistic data.

func (*Runtime) CloseOOBStatsFile 

func (rt *Runtime) CloseOOBStatsFile() (e error)

CloseOOBStatsFile will close statistics file for writing.

func (*Runtime) ComputeCM 

func (rt *Runtime) ComputeCM(sampleListID []int,
	vs, actuals, predicts []string,
) (
	cm *CM,
)

ComputeCM will compute confusion matrix of sample using value space, actual and prediction values.

func (*Runtime) ComputeStatFromCM 

func (rt *Runtime) ComputeStatFromCM(stat *Stat, cm *CM)

ComputeStatFromCM will compute statistic using confusion matrix.

func (*Runtime) ComputeStatTotal 

func (rt *Runtime) ComputeStatTotal(stat *Stat)

ComputeStatTotal compute total statistic.

func (*Runtime) Finalize 

func (rt *Runtime) Finalize() (e error)

Finalize finish the runtime, compute total statistic, write it to file, and close the file.

func (*Runtime) Initialize 

func (rt *Runtime) Initialize() error

Initialize will start the runtime for processing by saving start time and opening stats file.

func (*Runtime) OOBStats 

func (rt *Runtime) OOBStats() *Stats

OOBStats return all statistic objects.

func (*Runtime) OpenOOBStatsFile 

func (rt *Runtime) OpenOOBStatsFile() error

OpenOOBStatsFile will open statistic file for output.

func (*Runtime) Performance 

func (rt *Runtime) Performance(samples tabula.ClasetInterface,
	predicts []string, probs []float64,
) (
	perfs Stats,
)

Performance given an actuals class label and their probabilities, compute the performance statistic of classifier.

Algorithm, (1) Sort the probabilities in descending order. (2) Sort the actuals and predicts using sorted index from probs (3) Compute tpr, fpr, precision (4) Write performance to file.

func (*Runtime) PrintOobStat 

func (rt *Runtime) PrintOobStat(stat *Stat, cm *CM)

PrintOobStat will print the out-of-bag statistic to standard output.

func (*Runtime) PrintStat 

func (rt *Runtime) PrintStat(stat *Stat)

PrintStat will print statistic value to standard output.

func (*Runtime) PrintStatTotal 

func (rt *Runtime) PrintStatTotal(st *Stat)

PrintStatTotal will print total statistic to standard output.

func (*Runtime) StatTotal 

func (rt *Runtime) StatTotal() *Stat

StatTotal return total statistic.

func (*Runtime) WriteOOBStat 

func (rt *Runtime) WriteOOBStat(stat *Stat) error

WriteOOBStat will write statistic of process to file.

func (*Runtime) WritePerformance 

func (rt *Runtime) WritePerformance() error

WritePerformance will write performance data to file.

type Stat 

type Stat struct {
	// ID unique id for this statistic (e.g. number of tree).
	ID int64

	// StartTime contain the start time of classifier in unix timestamp.
	StartTime int64

	// EndTime contain the end time of classifier in unix timestamp.
	EndTime int64

	// ElapsedTime contain actual time, in seconds, between end and start
	// time.
	ElapsedTime int64

	// TP contain true-positive value.
	TP int64

	// FP contain false-positive value.
	FP int64

	// TN contain true-negative value.
	TN int64

	// FN contain false-negative value.
	FN int64

	// OobError contain out-of-bag error.
	OobError float64

	// OobErrorMean contain mean of out-of-bag error.
	OobErrorMean float64

	// TPRate contain true-positive rate (recall): tp/(tp+fn)
	TPRate float64

	// FPRate contain false-positive rate: fp/(fp+tn)
	FPRate float64

	// TNRate contain true-negative rate: tn/(tn+fp)
	TNRate float64

	// Precision contain: tp/(tp+fp)
	Precision float64

	// FMeasure contain value of F-measure or the harmonic mean of
	// precision and recall.
	FMeasure float64

	// Accuracy contain the degree of closeness of measurements of a
	// quantity to that quantity's true value.
	Accuracy float64

	// AUC contain the area under curve.
	AUC float64
}

Stat hold statistic value of classifier, including TP rate, FP rate, precision, and recall.

func (*Stat) End 

func (stat *Stat) End()

End will stop the timer and compute the elapsed time.

func (*Stat) Recall 

func (stat *Stat) Recall() float64

Recall return value of recall.

func (*Stat) SetAUC 

func (stat *Stat) SetAUC(v float64)

SetAUC will set the AUC value.

func (*Stat) SetFPRate 

func (stat *Stat) SetFPRate(fp, n int64)

SetFPRate will set FP and FPRate using number of negative `n`.

func (*Stat) SetPrecisionFromRate 

func (stat *Stat) SetPrecisionFromRate(p, n int64)

SetPrecisionFromRate will set Precision value using tprate and fprate. `p` and `n` is the number of positive and negative class in samples.

func (*Stat) SetTPRate 

func (stat *Stat) SetTPRate(tp, p int64)

SetTPRate will set TP and TPRate using number of positive `p`.

func (*Stat) Start 

func (stat *Stat) Start()

Start will start the timer.

func (*Stat) Sum 

func (stat *Stat) Sum(other *Stat)

Sum will add statistic from other stat object to current stat, not including the start and end time.

func (*Stat) ToRow 

func (stat *Stat) ToRow() (row *tabula.Row)

ToRow will convert the stat to tabula.row in the order of Stat field.

func (*Stat) Write 

func (stat *Stat) Write(file string) (e error)

Write will write the content of stat to `file`.

type Stats 

type Stats []*Stat

Stats define list of statistic values.

func (*Stats) Accuracies 

func (stats *Stats) Accuracies() (accuracies []float64)

Accuracies return all accuracy values.

func (*Stats) Add 

func (stats *Stats) Add(stat *Stat)

Add will add other stat object to the slice.

func (*Stats) EndTimes 

func (stats *Stats) EndTimes() (times []int64)

EndTimes return all end times in unix timestamp.

func (*Stats) FMeasures 

func (stats *Stats) FMeasures() (fmeasures []float64)

FMeasures return all F-measure values.

func (*Stats) FPRates 

func (stats *Stats) FPRates() (fprates []float64)

FPRates return all false-positive rate values.

func (*Stats) OobErrorMeans 

func (stats *Stats) OobErrorMeans() (oobmeans []float64)

OobErrorMeans return all out-of-bag error mean values.

func (*Stats) Precisions 

func (stats *Stats) Precisions() (precs []float64)

Precisions return all precision values.

func (*Stats) Recalls 

func (stats *Stats) Recalls() (recalls []float64)

Recalls return all recall values.

func (*Stats) StartTimes 

func (stats *Stats) StartTimes() (times []int64)

StartTimes return all start times in unix timestamp.

func (*Stats) TNRates 

func (stats *Stats) TNRates() (tnrates []float64)

TNRates will return all true-negative rate values.

func (*Stats) TPRates 

func (stats *Stats) TPRates() (tprates []float64)

TPRates return all true-positive rate values.

func (*Stats) Write 

func (stats *Stats) Write(file string) (e error)

Write will write all statistic data to `file`.

Source Files

View all Source files

Directories

Path Synopsis
Package cart implement the Classification and Regression Tree by Breiman, et al.
 Package cart implement the Classification and Regression Tree by Breiman, et al.
Package crf implement the cascaded random forest algorithm, proposed by Baumann et.al in their paper:
 Package crf implement the cascaded random forest algorithm, proposed by Baumann et.al in their paper:
Package rf implement ensemble of classifiers using random forest algorithm by Breiman and Cutler.
 Package rf implement ensemble of classifiers using random forest algorithm by Breiman and Cutler.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.