README.md

Gradient Descent Regressor for GoLang

This package allows you to create simple linear models with the Batch and Stochastic Gradient Descent methods.

Batch Gradient Descent

Sample usage for Batch Gradient Descent:

import "github.com/sschrs/GDRegressor"

x := [][]float64{{1.51}, {1.97}, {0.13}, {0.08}, {0.57}, {0.24}, {1.07}, {1.04}, {0.76}, {1.35}}
y := [][]float64{{9.08}, {10.64}, {4.73}, {5.16}, {5.66}, {4.93}, {7.21}, {7.13}, {7.33}, {5.94}}

bgd := GDRegressor.NewBGDRegressor()
bgd.Iterations = 1000
bgd.Eta = 0.1
bgd.Fit(x, y)

Make predictions:

predictedValues, err := bgd.Predict(x)
if err != nil {
    fmt.Println(err)
}
rmse := GDRegressor.RMSE(y, predictedValues)

Get intercept ant coefficients:

intercept := bgd.Intercept()
coef := bgd.Coef()

Stochastic Gradient Descent

Sample usage for Stochastic Gradient Descent:

import "github.com/sschrs/GDRegressor"
x := [][]float64{{1.51}, {1.97}, {0.13}, {0.08}, {0.57}, {0.24}, {1.07}, {1.04}, {0.76}, {1.35}}
y := [][]float64{{9.08}, {10.64}, {4.73}, {5.16}, {5.66}, {4.93}, {7.21}, {7.13}, {7.33}, {5.94}}

sgd := GDRegressor.NewSGDRegressor()
sgd.Iterations = 1000
sgd.Eta = 0.1
sgd.Fit(x, y)

Make predictions:

predictedValues, err := sgd.Predict(x)
if err != nil {
    fmt.Println(err)
}
rmse := GDRegressor.RMSE(y, predictedValues)

Get intercept ant coefficients:

intercept := sgd.Intercept()
coef := sgd.Coef()