func readShuffled(in io.Reader, rand *rand.Rand) []TrainingInstance {
reader, err := input.NewTrainDataReader(in)
common.ExitIfError("Error reading data: ", err)
instances := make([]TrainingInstance, 0)
for {
err := reader.Scan()
if err == io.EOF {
break
} else {
common.ExitIfError("Error reading data: ", err)
}
instance := TrainingInstance{
X: reader.InputVector().Dup(),
Y: reader.Label(),
}
idx := rand.Int63n(int64(len(instances)) + 1)
if int(idx) == len(instances) {
instances = append(instances, instance)
} else {
instances = append(instances, instances[idx])
instances[idx] = instance
}
}
return instances
}
func processData(filename string, fun dataFun) error {
f, err := os.Open(filename)
if err != nil {
return err
}
bar, err := common.CreateFileProgress(f)
defer bar.Finish()
bar.Start()
reader, err := input.NewTrainDataReader(bar.NewProxyReader(bufio.NewReader(f)))
if err != nil {
return err
}
for {
if err := reader.Scan(); err == io.EOF {
break
} else if err != nil {
return err
}
if err := fun(reader.Label(), reader.InputVector()); err != nil {
return err
}
}
return nil
}
func main() {
flag.Parse()
if flag.NArg() != 1 {
flag.Usage()
os.Exit(1)
}
f, err := os.Open(flag.Arg(0))
common.ExitIfError("Could not open data file: ", err)
defer f.Close()
reader, err := input.NewTrainDataReader(f)
common.ExitIfError("Error reading data: ", err)
for {
err := reader.Scan()
if err == io.EOF {
break
} else {
common.ExitIfError("Error reading data: ", err)
}
if *features {
fmt.Printf("%d %s\n", reader.Label(), floatSliceToString(reader.InputVector().Layer(addr.FEATURE)))
} else {
fmt.Printf("%d %s\n", reader.Label(), floatSliceToString(reader.InputVector().All()))
}
}
}
