但是集群spark1.0.2平臺已搭好，不方便再升級了。所以我們最后的方法是利用對象序列化儲存搞定了[存儲model]:
val serial_out = new ObjectOutputStream(new FileOutputStream("svm_model.obj"))
serial_out.writeObject(model)
serial_out.close()
[加載model:]
val serial_in = new ObjectInputStream(new FileInputStream("svm_model.obj"))
val saved_model = serial_in.readObject().asInstanceOf[SVMModel]
