如何根据其类型创建sparkml对象

Question

在代码中，我有一个SparkML PipelineModel，可以作为最后一个阶段，并预测一个RandomForest或GradientTreeBoosting模型。我想根据类型创建对象。我已经尝试过了，但是由于变量被包含在if块内，所以它不起作用。无论如何，有没有投射正确的对象，还是我需要将vectors循环封装在每个if块内？

    if (object instanceof PipelineModel) {
        PipelineModel pipelineModel = (PipelineModel) object;
        if (pipelineModel.stages()[pipelineModel.stages().length - 1] instanceof RandomForestRegressionModel) {
            RandomForestRegressionModel rfModel = (RandomForestRegressionModel) pipelineModel.stages()[pipelineModel.stages().length - 1];
        }
        if (pipelineModel.stages()[pipelineModel.stages().length - 1] instanceof GBTRegressionModel) {
            GBTRegressionModel rfModel = (GBTRegressionModel) pipelineModel.stages()[pipelineModel.stages().length - 1];
        }
        for (int i = 0; i < vectors.size(); i++) {
            double[] vector = new double[vectors.get(i).length];
            for (int j = vector.length; j < vector.length; j++) {
                vector[j] = vectors.get(i)[j];
            }
            double predictResult = rfModel.predict(Vectors.dense(vector));
            result.add(predictResult);
        }
    }