var loop int

var file string

var mode string

var model string

var lambda float64

var verbose bool

flag.IntVar(&loop, "l", 10, "number of iterations")

flag.Float64Var(&lambda, "c", 0.1, "regularization parameter")

flag.StringVar(&file, "f", "", "data file")

flag.StringVar(&mode, "m", "", "mode {learn, test}")

flag.StringVar(&model, "w", "", "model file")

flag.BoolVar(&verbose, "v", false, "verbose mode")

flag.Parse()

if file == "" {

panic("Data must be specified")

}

if mode == "learn" {

flag.Parse()

X, y := LoadFromFile(file)

p := NewAdaGrad(lambda, loop)

p.Fit(X, y)

SaveModel(p, model)

} else if mode == "test" {

p := LoadModel(model)

X_test, y_test := LoadFromFile(file)

num_corr := 0.

n := 0.

pred_y := []string{}

for i, X_i := range X_test {

pred_y_i := p.Predict(X_i)

pred_y = append(pred_y, pred_y_i)

if verbose {

fmt.Println(pred_y_i)

}

if pred_y_i == y_test[i] {

num_corr += 1

}

n += 1

}

acc := num_corr / n

confusionMatrix(y_test, pred_y)

fmt.Println("Acc:", acc)

} else {

panic("Invalid mode")

}