多層感知器分類器（MLPC）是基於前饋人工神經網路(ANN)的分類器。 MLPC由多個節點層組成。 每個層完全連線到網路中的下一層。 輸入層中的節點表示輸入資料。 所有其他節點,通過輸入與節點的權重w和偏置b的線性組合,並應用啟用函式,將輸入對映到輸出。 對於具有K + 1層的MLPC，這可以以矩陣形式寫成如下：
中間層中的節點使用sigmoid（logistic）函式：
輸出層中的節點使用softmax函式：
輸出層中的節點數量N對應於類的數量。
MLPC採用反向傳播學習模型(BP演算法)。 我們使用用於優化的邏輯損失函式和L-BFGS作為優化程式。
匯入包
import org.apache.spark.sql.SparkSession import org.apache.spark.sql.Dataset import org.apache.spark.sql.Row import org.apache.spark.sql.DataFrame import org.apache.spark.sql.Column import org.apache.spark.sql.DataFrameReader import org.apache.spark.rdd.RDD import org.apache.spark.sql.catalyst.encoders.ExpressionEncoder import org.apache.spark.sql.Encoder import org.apache.spark.sql.DataFrameStatFunctions import org.apache.spark.sql.functions._ import org.apache.spark.ml.linalg.Vectors import org.apache.spark.ml.feature.VectorAssembler import org.apache.spark.ml.classification.MultilayerPerceptronClassifier import org.apache.spark.ml.evaluation.MulticlassClassificationEvaluator
匯入資料來源
val spark = SparkSession.builder().appName("Spark Multilayer perceptron classifier").config("spark.some.config.option", "some-value").getOrCreate()
// For implicit conversions like converting RDDs to DataFrames
import spark.implicits._
val dataList: List[(Double, String, Double, Double, String, Double, Double, Double, Double)] = List(
(0, "male", 37, 10, "no", 3, 18, 7, 4),
(0, "female", 27, 4, "no", 4, 14, 6, 4),
(0, "female", 32, 15, "yes", 1, 12, 1, 4),
(0, "male", 57, 15, "yes", 5, 18, 6, 5),
(0, "male", 22, 0.75, "no", 2, 17, 6, 3),
(0, "female", 32, 1.5, "no", 2, 17, 5, 5),
(0, "female", 22, 0.75, "no", 2, 12, 1, 3),
(0, "male", 57, 15, "yes", 2, 14, 4, 4),
(0, "female", 32, 15, "yes", 4, 16, 1, 2),
(0, "male", 22, 1.5, "no", 4, 14, 4, 5),
(0, "male", 37, 15, "yes", 2, 20, 7, 2),
(0, "male", 27, 4, "yes", 4, 18, 6, 4),
(0, "male", 47, 15, "yes", 5, 17, 6, 4),
(0, "female", 22, 1.5, "no", 2, 17, 5, 4),
(0, "female", 27, 4, "no", 4, 14, 5, 4),
(0, "female", 37, 15, "yes", 1, 17, 5, 5),
(0, "female", 37, 15, "yes", 2, 18, 4, 3),
(0, "female", 22, 0.75, "no", 3, 16, 5, 4),
(0, "female", 22, 1.5, "no", 2, 16, 5, 5),
(0, "female", 27, 10, "yes", 2, 14, 1, 5),
(0, "female", 22, 1.5, "no", 2, 16, 5, 5),
(0, "female", 22, 1.5, "no", 2, 16, 5, 5),
(0, "female", 27, 10, "yes", 4, 16, 5, 4),
(0, "female", 32, 10, "yes", 3, 14, 1, 5),
(0, "male", 37, 4, "yes", 2, 20, 6, 4),
(0, "female", 22, 1.5, "no", 2, 18, 5, 5),
(0, "female", 27, 7, "no", 4, 16, 1, 5),
(0, "male", 42, 15, "yes", 5, 20, 6, 4),
(0, "male", 27, 4, "yes", 3, 16, 5, 5),
(0, "female", 27, 4, "yes", 3, 17, 5, 4),
(0, "male", 42, 15, "yes", 4, 20, 6, 3),
(0, "female", 22, 1.5, "no", 3, 16, 5, 5),
(0, "male", 27, 0.417, "no", 4, 17, 6, 4),
(0, "female", 42, 15, "yes", 5, 14, 5, 4),
(0, "male", 32, 4, "yes", 1, 18, 6, 4),
(0, "female", 22, 1.5, "no", 4, 16, 5, 3),
(0, "female", 42, 15, "yes", 3, 12, 1, 4),
(0, "female", 22, 4, "no", 4, 17, 5, 5),
(0, "male", 22, 1.5, "yes", 1, 14, 3, 5),
(0, "female", 22, 0.75, "no", 3, 16, 1, 5),
(0, "male", 32, 10, "yes", 5, 20, 6, 5),
(0, "male", 52, 15, "yes", 5, 18, 6, 3),
(0, "female", 22, 0.417, "no", 5, 14, 1, 4),
(0, "female", 27, 4, "yes", 2, 18, 6, 1),
(0, "female", 32, 7, "yes", 5, 17, 5, 3),
(0, "male", 22, 4, "no", 3, 16, 5, 5),
(0, "female", 27, 7, "yes", 4, 18, 6, 5),
(0, "female", 42, 15, "yes", 2, 18, 5, 4),
(0, "male", 27, 1.5, "yes", 4, 16, 3, 5),
(0, "male", 42, 15, "yes", 2, 20, 6, 4),
(0, "female", 22, 0.75, "no", 5, 14, 3, 5),
(0, "male", 32, 7, "yes", 2, 20, 6, 4),
(0, "male", 27, 4, "yes", 5, 20, 6, 5),
(0, "male", 27, 10, "yes", 4, 20, 6, 4),
(0, "male", 22, 4, "no", 1, 18, 5, 5),
(0, "female", 37, 15, "yes", 4, 14, 3, 1),
(0, "male", 22, 1.5, "yes", 5, 16, 4, 4),
(0, "female", 37, 15, "yes", 4, 17, 1, 5),
(0, "female", 27, 0.75, "no", 4, 17, 5, 4),
(0, "male", 32, 10, "yes", 4, 20, 6, 4),
(0, "female", 47, 15, "yes", 5, 14, 7, 2),
(0, "male", 37, 10, "yes", 3, 20, 6, 4),
(0, "female", 22, 0.75, "no", 2, 16, 5, 5),
(0, "male", 27, 4, "no", 2, 18, 4, 5),
(0, "male", 32, 7, "no", 4, 20, 6, 4),
(0, "male", 42, 15, "yes", 2, 17, 3, 5),
(0, "male", 37, 10, "yes", 4, 20, 6, 4),
(0, "female", 47, 15, "yes", 3, 17, 6, 5),
(0, "female", 22, 1.5, "no", 5, 16, 5, 5),
(0, "female", 27, 1.5, "no", 2, 16, 6, 4),
(0, "female", 27, 4, "no", 3, 17, 5, 5),
(0, "female", 32, 10, "yes", 5, 14, 4, 5),
(0, "female", 22, 0.125, "no", 2, 12, 5, 5),
(0, "male", 47, 15, "yes", 4, 14, 4, 3),
(0, "male", 32, 15, "yes", 1, 14, 5, 5),
(0, "male", 27, 7, "yes", 4, 16, 5, 5),
(0, "female", 22, 1.5, "yes", 3, 16, 5, 5),
(0, "male", 27, 4, "yes", 3, 17, 6, 5),
(0, "female", 22, 1.5, "no", 3, 16, 5, 5),
(0, "male", 57, 15, "yes", 2, 14, 7, 2),
(0, "male", 17.5, 1.5, "yes", 3, 18, 6, 5),
(0, "male", 57, 15, "yes", 4, 20, 6, 5),
(0, "female", 22, 0.75, "no", 2, 16, 3, 4),
(0, "male", 42, 4, "no", 4, 17, 3, 3),
(0, "female", 22, 1.5, "yes", 4, 12, 1, 5),
(0, "female", 22, 0.417, "no", 1, 17, 6, 4),
(0, "female", 32, 15, "yes", 4, 17, 5, 5),
(0, "female", 27, 1.5, "no", 3, 18, 5, 2),
(0, "female", 22, 1.5, "yes", 3, 14, 1, 5),
(0, "female", 37, 15, "yes", 3, 14, 1, 4),
(0, "female", 32, 15, "yes", 4, 14, 3, 4),
(0, "male", 37, 10, "yes", 2, 14, 5, 3),
(0, "male", 37, 10, "yes", 4, 16, 5, 4),
(0, "male", 57, 15, "yes", 5, 20, 5, 3),
(0, "male", 27, 0.417, "no", 1, 16, 3, 4),
(0, "female", 42, 15, "yes", 5, 14, 1, 5),
(0, "male", 57, 15, "yes", 3, 16, 6, 1),
(0, "male", 37, 10, "yes", 1, 16, 6, 4),
(0, "male", 37, 15, "yes", 3, 17, 5, 5),
(0, "male", 37, 15, "yes", 4, 20, 6, 5),
(0, "female", 27, 10, "yes", 5, 14, 1, 5),
(0, "male", 37, 10, "yes", 2, 18, 6, 4),
(0, "female", 22, 0.125, "no", 4, 12, 4, 5),
(0, "male", 57, 15, "yes", 5, 20, 6, 5),
(0, "female", 37, 15, "yes", 4, 18, 6, 4),
(0, "male", 22, 4, "yes", 4, 14, 6, 4),
(0, "male", 27, 7, "yes", 4, 18, 5, 4),
(0, "male", 57, 15, "yes", 4, 20, 5, 4),
(0, "male", 32, 15, "yes", 3, 14, 6, 3),
(0, "female", 22, 1.5, "no", 2, 14, 5, 4),
(0, "female", 32, 7, "yes", 4, 17, 1, 5),
(0, "female", 37, 15, "yes", 4, 17, 6, 5),
(0, "female", 32, 1.5, "no", 5, 18, 5, 5),
(0, "male", 42, 10, "yes", 5, 20, 7, 4),
(0, "female", 27, 7, "no", 3, 16, 5, 4),
(0, "male", 37, 15, "no", 4, 20, 6, 5),
(0, "male", 37, 15, "yes", 4, 14, 3, 2),
(0, "male", 32, 10, "no", 5, 18, 6, 4),
(0, "female", 22, 0.75, "no", 4, 16, 1, 5),
(0, "female", 27, 7, "yes", 4, 12, 2, 4),
(0, "female", 27, 7, "yes", 2, 16, 2, 5),
(0, "female", 42, 15, "yes", 5, 18, 5, 4),
(0, "male", 42, 15, "yes", 4, 17, 5, 3),
(0, "female", 27, 7, "yes", 2, 16, 1, 2),
(0, "female", 22, 1.5, "no", 3, 16, 5, 5),
(0, "male", 37, 15, "yes", 5, 20, 6, 5),
(0, "female", 22, 0.125, "no", 2, 14, 4, 5),
(0, "male", 27, 1.5, "no", 4, 16, 5, 5),
(0, "male", 32, 1.5, "no", 2, 18, 6, 5),
(0, "male", 27, 1.5, "no", 2, 17, 6, 5),
(0, "female", 27, 10, "yes", 4, 16, 1, 3),
(0, "male", 42, 15, "yes", 4, 18, 6, 5),
(0, "female", 27, 1.5, "no", 2, 16, 6, 5),
(0, "male", 27, 4, "no", 2, 18, 6, 3),
(0, "female", 32, 10, "yes", 3, 14, 5, 3),
(0, "female", 32, 15, "yes", 3, 18, 5, 4),
(0, "female", 22, 0.75, "no", 2, 18, 6, 5),
(0, "female", 37, 15, "yes", 2, 16, 1, 4),
(0, "male", 27, 4, "yes", 4, 20, 5, 5),
(0, "male", 27, 4, "no", 1, 20, 5, 4),
(0, "female", 27, 10, "yes", 2, 12, 1, 4),
(0, "female", 32, 15, "yes", 5, 18, 6, 4),
(0, "male", 27, 7, "yes", 5, 12, 5, 3),
(0, "male", 52, 15, "yes", 2, 18, 5, 4),
(0, "male", 27, 4, "no", 3, 20, 6, 3),
(0, "male", 37, 4, "yes", 1, 18, 5, 4),
(0, "male", 27, 4, "yes", 4, 14, 5, 4),
(0, "female", 52, 15, "yes", 5, 12, 1, 3),
(0, "female", 57, 15, "yes", 4, 16, 6, 4),
(0, "male", 27, 7, "yes", 1, 16, 5, 4),
(0, "male", 37, 7, "yes", 4, 20, 6, 3),
(0, "male", 22, 0.75, "no", 2, 14, 4, 3),
(0, "male", 32, 4, "yes", 2, 18, 5, 3),
(0, "male", 37, 15, "yes", 4, 20, 6, 3),
(0, "male", 22, 0.75, "yes", 2, 14, 4, 3),
(0, "male", 42, 15, "yes", 4, 20, 6, 3),
(0, "female", 52, 15, "yes", 5, 17, 1, 1),
(0, "female", 37, 15, "yes", 4, 14, 1, 2),
(0, "male", 27, 7, "yes", 4, 14, 5, 3),
(0, "male", 32, 4, "yes", 2, 16, 5, 5),
(0, "female", 27, 4, "yes", 2, 18, 6, 5),
(0, "female", 27, 4, "yes", 2, 18, 5, 5),
(0, "male", 37, 15, "yes", 5, 18, 6, 5),
(0, "female", 47, 15, "yes", 5, 12, 5, 4),
(0, "female", 32, 10, "yes", 3, 17, 1, 4),
(0, "female", 27, 1.5, "yes", 4, 17, 1, 2),
(0, "female", 57, 15, "yes", 2, 18, 5, 2),
(0, "female", 22, 1.5, "no", 4, 14, 5, 4),
(0, "male", 42, 15, "yes", 3, 14, 3, 4),
(0, "male", 57, 15, "yes", 4, 9, 2, 2),
(0, "male", 57, 15, "yes", 4, 20, 6, 5),
(0, "female", 22, 0.125, "no", 4, 14, 4, 5),
(0, "female", 32, 10, "yes", 4, 14, 1, 5),
(0, "female", 42, 15, "yes", 3, 18, 5, 4),
(0, "female", 27, 1.5, "no", 2, 18, 6, 5),
(0, "male", 32, 0.125, "yes", 2, 18, 5, 2),
(0, "female", 27, 4, "no", 3, 16, 5, 4),
(0, "female", 27, 10, "yes", 2, 16, 1, 4),
(0, "female", 32, 7, "yes", 4, 16, 1, 3),
(0, "female", 37, 15, "yes", 4, 14, 5, 4),
(0, "female", 42, 15, "yes", 5, 17, 6, 2),
(0, "male", 32, 1.5, "yes", 4, 14, 6, 5),
(0, "female", 32, 4, "yes", 3, 17, 5, 3),
(0, "female", 37, 7, "no", 4, 18, 5, 5),
(0, "female", 22, 0.417, "yes", 3, 14, 3, 5),
(0, "female", 27, 7, "yes", 4, 14, 1, 5),
(0, "male", 27, 0.75, "no", 3, 16, 5, 5),
(0, "male", 27, 4, "yes", 2, 20, 5, 5),
(0, "male", 32, 10, "yes", 4, 16, 4, 5),
(0, "male", 32, 15, "yes", 1, 14, 5, 5),
(0, "male", 22, 0.75, "no", 3, 17, 4, 5),
(0, "female", 27, 7, "yes", 4, 17, 1, 4),
(0, "male", 27, 0.417, "yes", 4, 20, 5, 4),
(0, "male", 37, 15, "yes", 4, 20, 5, 4),
(0, "female", 37, 15, "yes", 2, 14, 1, 3),
(0, "male", 22, 4, "yes", 1, 18, 5, 4),
(0, "male", 37, 15, "yes", 4, 17, 5, 3),
(0, "female", 22, 1.5, "no", 2, 14, 4, 5),
(0, "male", 52, 15, "yes", 4, 14, 6, 2),
(0, "female", 22, 1.5, "no", 4, 17, 5, 5),
(0, "male", 32, 4, "yes", 5, 14, 3, 5),
(0, "male", 32, 4, "yes", 2, 14, 3, 5),
(0, "female", 22, 1.5, "no", 3, 16, 6, 5),
(0, "male", 27, 0.75, "no", 2, 18, 3, 3),
(0, "female", 22, 7, "yes", 2, 14, 5, 2),
(0, "female", 27, 0.75, "no", 2, 17, 5, 3),
(0, "female", 37, 15, "yes", 4, 12, 1, 2),
(0, "female", 22, 1.5, "no", 1, 14, 1, 5),
(0, "female", 37, 10, "no", 2, 12, 4, 4),
(0, "female", 37, 15, "yes", 4, 18, 5, 3),
(0, "female", 42, 15, "yes", 3, 12, 3, 3),
(0, "male", 22, 4, "no", 2, 18, 5, 5),
(0, "male", 52, 7, "yes", 2, 20, 6, 2),
(0, "male", 27, 0.75, "no", 2, 17, 5, 5),
(0, "female", 27, 4, "no", 2, 17, 4, 5),
(0, "male", 42, 1.5, "no", 5, 20, 6, 5),
(0, "male", 22, 1.5, "no", 4, 17, 6, 5),
(0, "male", 22, 4, "no", 4, 17, 5, 3),
(0, "female", 22, 4, "yes", 1, 14, 5, 4),
(0, "male", 37, 15, "yes", 5, 20, 4, 5),
(0, "female", 37, 10, "yes", 3, 16, 6, 3),
(0, "male", 42, 15, "yes", 4, 17, 6, 5),
(0, "female", 47, 15, "yes", 4, 17, 5, 5),
(0, "male", 22, 1.5, "no", 4, 16, 5, 4),
(0, "female", 32, 10, "yes", 3, 12, 1, 4),
(0, "female", 22, 7, "yes", 1, 14, 3, 5),
(0, "female", 32, 10, "yes", 4, 17, 5, 4),
(0, "male", 27, 1.5, "yes", 2, 16, 2, 4),
(0, "male", 37, 15, "yes", 4, 14, 5, 5),
(0, "male", 42, 4, "yes", 3, 14, 4, 5),
(0, "female", 37, 15, "yes", 5, 14, 5, 4),
(0, "female", 32, 7, "yes", 4, 17, 5, 5),
(0, "female", 42, 15, "yes", 4, 18, 6, 5),
(0, "male", 27, 4, "no", 4, 18, 6, 4),
(0, "male", 22, 0.75, "no", 4, 18, 6, 5),
(0, "male", 27, 4, "yes", 4, 14, 5, 3),
(0, "female", 22, 0.75, "no", 5, 18, 1, 5),
(0, "female", 52, 15, "yes", 5, 9, 5, 5),
(0, "male", 32, 10, "yes", 3, 14, 5, 5),
(0, "female", 37, 15, "yes", 4, 16, 4, 4),
(0, "male", 32, 7, "yes", 2, 20, 5, 4),
(0, "female", 42, 15, "yes", 3, 18, 1, 4),
(0, "male", 32, 15, "yes", 1, 16, 5, 5),
(0, "male", 27, 4, "yes", 3, 18, 5, 5),
(0, "female", 32, 15, "yes", 4, 12, 3, 4),
(0, "male", 22, 0.75, "yes", 3, 14, 2, 4),
(0, "female", 22, 1.5, "no", 3, 16, 5, 3),
(0, "female", 42, 15, "yes", 4, 14, 3, 5),
(0, "female", 52, 15, "yes", 3, 16, 5, 4),
(0, "male", 37, 15, "yes", 5, 20, 6, 4),
(0, "female", 47, 15, "yes", 4, 12, 2, 3),
(0, "male", 57, 15, "yes", 2, 20, 6, 4),
(0, "male", 32, 7, "yes", 4, 17, 5, 5),
(0, "female", 27, 7, "yes", 4, 17, 1, 4),
(0, "male", 22, 1.5, "no", 1, 18, 6, 5),
(0, "female", 22, 4, "yes", 3, 9, 1, 4),
(0, "female", 22, 1.5, "no", 2, 14, 1, 5),
(0, "male", 42, 15, "yes", 2, 20, 6, 4),
(0, "male", 57, 15, "yes", 4, 9, 2, 4),
(0, "female", 27, 7, "yes", 2, 18, 1, 5),
(0, "female", 22, 4, "yes", 3, 14, 1, 5),
(0, "male", 37, 15, "yes", 4, 14, 5, 3),
(0, "male", 32, 7, "yes", 1, 18, 6, 4),
(0, "female", 22, 1.5, "no", 2, 14, 5, 5),
(0, "female", 22, 1.5, "yes", 3, 12, 1, 3),
(0, "male", 52, 15, "yes", 2, 14, 5, 5),
(0, "female", 37, 15, "yes", 2, 14, 1, 1),
(0, "female", 32, 10, "yes", 2, 14, 5, 5),
(0, "male", 42, 15, "yes", 4, 20, 4, 5),
(0, "female", 27, 4, "yes", 3, 18, 4, 5),
(0, "male", 37, 15, "yes", 4, 20, 6, 5),
(0, "male", 27, 1.5, "no", 3, 18, 5, 5),
(0, "female", 22, 0.125, "no", 2, 16, 6, 3),
(0, "male", 32, 10, "yes", 2, 20, 6, 3),
(0, "female", 27, 4, "no", 4, 18, 5, 4),
(0, "female", 27, 7, "yes", 2, 12, 5, 1),
(0, "male", 32, 4, "yes", 5, 18, 6, 3),
(0, "female", 37, 15, "yes", 2, 17, 5, 5),
(0, "male", 47, 15, "no", 4, 20, 6, 4),
(0, "male", 27, 1.5, "no", 1, 18, 5, 5),
(0, "male", 37, 15, "yes", 4, 20, 6, 4),
(0, "female", 32, 15, "yes", 4, 18, 1, 4),
(0, "female", 32, 7, "yes", 4, 17, 5, 4),
(0, "female", 42, 15, "yes", 3, 14, 1, 3),
(0, "female", 27, 7, "yes", 3, 16, 1, 4),
(0, "male", 27, 1.5, "no", 3, 16, 4, 2),
(0, "male", 22, 1.5, "no", 3, 16, 3, 5),
(0, "male", 27, 4, "yes", 3, 16, 4, 2),
(0, "female", 27, 7, "yes", 3, 12, 1, 2),
(0, "female", 37, 15, "yes", 2, 18, 5, 4),
(0, "female", 37, 7, "yes", 3, 14, 4, 4),
(0, "male", 22, 1.5, "no", 2, 16, 5, 5),
(0, "male", 37, 15, "yes", 5, 20, 5, 4),
(0, "female", 22, 1.5, "no", 4, 16, 5, 3),
(0, "female", 32, 10, "yes", 4, 16, 1, 5),
(0, "male", 27, 4, "no", 2, 17, 5, 3),
(0, "female", 22, 0.417, "no", 4, 14, 5, 5),
(0, "female", 27, 4, "no", 2, 18, 5, 5),
(0, "male", 37, 15, "yes", 4, 18, 5, 3),
(0, "male", 37, 10, "yes", 5, 20, 7, 4),
(0, "female", 27, 7, "yes", 2, 14, 4, 2),
(0, "male", 32, 4, "yes", 2, 16, 5, 5),
(0, "male", 32, 4, "yes", 2, 16, 6, 4),
(0, "male", 22, 1.5, "no", 3, 18, 4, 5),
(0, "female", 22, 4, "yes", 4, 14, 3, 4),
(0, "female", 17.5, 0.75, "no", 2, 18, 5, 4),
(0, "male", 32, 10, "yes", 4, 20, 4, 5),
(0, "female", 32, 0.75, "no", 5, 14, 3, 3),
(0, "male", 37, 15, "yes", 4, 17, 5, 3),
(0, "male", 32, 4, "no", 3, 14, 4, 5),
(0, "female", 27, 1.5, "no", 2, 17, 3, 2),
(0, "female", 22, 7, "yes", 4, 14, 1, 5),
(0, "male", 47, 15, "yes", 5, 14, 6, 5),
(0, "male", 27, 4, "yes", 1, 16, 4, 4),
(0, "female", 37, 15, "yes", 5, 14, 1, 3),
(0, "male", 42, 4, "yes", 4, 18, 5, 5),
(0, "female", 32, 4, "yes", 2, 14, 1, 5),
(0, "male", 52, 15, "yes", 2, 14, 7, 4),
(0, "female", 22, 1.5, "no", 2, 16, 1, 4),
(0, "male", 52, 15, "yes", 4, 12, 2, 4),
(0, "female", 22, 0.417, "no", 3, 17, 1, 5),
(0, "female", 22, 1.5, "no", 2, 16, 5, 5),
(0, "male", 27, 4, "yes", 4, 20, 6, 4),
(0, "female", 32, 15, "yes", 4, 14, 1, 5),
(0, "female", 27, 1.5, "no", 2, 16, 3, 5),
(0, "male", 32, 4, "no", 1, 20, 6, 5),
(0, "male", 37, 15, "yes", 3, 20, 6, 4),
(0, "female", 32, 10, "no", 2, 16, 6, 5),
(0, "female", 32, 10, "yes", 5, 14, 5, 5),
(0, "male", 37, 1.5, "yes", 4, 18, 5, 3),
(0, "male", 32, 1.5, "no", 2, 18, 4, 4),
(0, "female", 32, 10, "yes", 4, 14, 1, 4),
(0, "female", 47, 15, "yes", 4, 18, 5, 4),
(0, "female", 27, 10, "yes", 5, 12, 1, 5),
(0, "male", 27, 4, "yes", 3, 16, 4, 5),
(0, "female", 37, 15, "yes", 4, 12, 4, 2),
(0, "female", 27, 0.75, "no", 4, 16, 5, 5),
(0, "female", 37, 15, "yes", 4, 16, 1, 5),
(0, "female", 32, 15, "yes", 3, 16, 1, 5),
(0, "female", 27, 10, "yes", 2, 16, 1, 5),
(0, "male", 27, 7, "no", 2, 20, 6, 5),
(0, "female", 37, 15, "yes", 2, 14, 1, 3),
(0, "male", 27, 1.5, "yes", 2, 17, 4, 4),
(0, "female", 22, 0.75, "yes", 2, 14, 1, 5),
(0, "male", 22, 4, "yes", 4, 14, 2, 4),
(0, "male", 42, 0.125, "no", 4, 17, 6, 4),
(0, "male", 27, 1.5, "yes", 4, 18, 6, 5),
(0, "male", 27, 7, "yes", 3, 16, 6, 3),
(0, "female", 52, 15, "yes", 4, 14, 1, 3),
(0, "male", 27, 1.5, "no", 5, 20, 5, 2),
(0, "female", 27, 1.5, "no", 2, 16, 5, 5),
(0, "female", 27, 1.5, "no", 3, 17, 5, 5),
(0, "male", 22, 0.125, "no", 5, 16, 4, 4),
(0, "female", 27, 4, "yes", 4, 16, 1, 5),
(0, "female", 27, 4, "yes", 4, 12, 1, 5),
(0, "female", 47, 15, "yes", 2, 14, 5, 5),
(0, "female", 32, 15, "yes", 3, 14, 5, 3),
(0, "male", 42, 7, "yes", 2, 16, 5, 5),
(0, "male", 22, 0.75, "no", 4, 16, 6, 4),
(0, "male", 27, 0.125, "no", 3, 20, 6, 5),
(0, "male", 32, 10, "yes", 3, 20, 6, 5),
(0, "female", 22, 0.417, "no", 5, 14, 4, 5),
(0, "female", 47, 15, "yes", 5, 14, 1, 4),
(0, "female", 32, 10, "yes", 3, 14, 1, 5),
(0, "male", 57, 15, "yes", 4, 17, 5, 5),
(0, "male", 27, 4, "yes", 3, 20, 6, 5),
(0, "female", 32, 7, "yes", 4, 17, 1, 5),
(0, "female", 37, 10, "yes", 4, 16, 1, 5),
(0, "female", 32, 10, "yes", 1, 18, 1, 4),
(0, "female", 22, 4, "no", 3, 14, 1, 4),
(0, "female", 27, 7, "yes", 4, 14, 3, 2),
(0, "male", 57, 15, "yes", 5, 18, 5, 2),
(0, "male", 32, 7, "yes", 2, 18, 5, 5),
(0, "female", 27, 1.5, "no", 4, 17, 1, 3),
(0, "male", 22, 1.5, "no", 4, 14, 5, 5),
(0, "female", 22, 1.5, "yes", 4, 14, 5, 4),
(0, "female", 32, 7, "yes", 3, 16, 1, 5),
(0, "female", 47, 15, "yes", 3, 16, 5, 4),
(0, "female", 22, 0.75, "no", 3, 16, 1, 5),
(0, "female", 22, 1.5, "yes", 2, 14, 5, 5),
(0, "female", 27, 4, "yes", 1, 16, 5, 5),
(0, "male", 52, 15, "yes", 4, 16, 5, 5),
(0, "male", 32, 10, "yes", 4, 20, 6, 5),
(0, "male", 47, 15, "yes", 4, 16, 6, 4),
(0, "female", 27, 7, "yes", 2, 14, 1, 2),
(0, "female", 22, 1.5, "no", 4, 14, 4, 5),
(0, "female", 32, 10, "yes", 2, 16, 5, 4),
(0, "female", 22, 0.75, "no", 2, 16, 5, 4),
(0, "female", 22, 1.5, "no", 2, 16, 5, 5),
(0, "female", 42, 15, "yes", 3, 18, 6, 4),
(0, "female", 27, 7, "yes", 5, 14, 4, 5),
(0, "male", 42, 15, "yes", 4, 16, 4, 4),
(0, "female", 57, 15, "yes", 3, 18, 5, 2),
(0, "male", 42, 15, "yes", 3, 18, 6, 2),
(0, "female", 32, 7, "yes", 2, 14, 1, 2),
(0, "male", 22, 4, "no", 5, 12, 4, 5),
(0, "female", 22, 1.5, "no", 1, 16, 6, 5),
(0, "female", 22, 0.75, "no", 1, 14, 4, 5),
(0, "female", 32, 15, "yes", 4, 12, 1, 5),
(0, "male", 22, 1.5, "no", 2, 18, 5, 3),
(0, "male", 27, 4, "yes", 5, 17, 2, 5),
(0, "female", 27, 4, "yes", 4, 12, 1, 5),
(0, "male", 42, 15, "yes", 5, 18, 5, 4),
(0, "male", 32, 1.5, "no", 2, 20, 7, 3),
(0, "male", 57, 15, "no", 4, 9, 3, 1),
(0, "male", 37, 7, "no", 4, 18, 5, 5),
(0, "male", 52, 15, "yes", 2, 17, 5, 4),
(0, "male", 47, 15, "yes", 4, 17, 6, 5),
(0, "female", 27, 7, "no", 2, 17, 5, 4),
(0, "female", 27, 7, "yes", 4, 14, 5, 5),
(0, "female", 22, 4, "no", 2, 14, 3, 3),
(0, "male", 37, 7, "yes", 2, 20, 6, 5),
(0, "male", 27, 7, "no", 4, 12, 4, 3),
(0, "male", 42, 10, "yes", 4, 18, 6, 4),
(0, "female", 22, 1.5, "no", 3, 14, 1, 5),
(0, "female", 22, 4, "yes", 2, 14, 1, 3),
(0, "female", 57, 15, "no", 4, 20, 6, 5),
(0, "male", 37, 15, "yes", 4, 14, 4, 3),
(0, "female", 27, 7, "yes", 3, 18, 5, 5),
(0, "female", 17.5, 10, "no", 4, 14, 4, 5),
(0, "male", 22, 4, "yes", 4, 16, 5, 5),
(0, "female", 27, 4, "yes", 2, 16, 1, 4),
(0, "female", 37, 15, "yes", 2, 14, 5, 1),
(0, "female", 22, 1.5, "no", 5, 14, 1, 4),
(0, "male", 27, 7, "yes", 2, 20, 5, 4),
(0, "male", 27, 4, "yes", 4, 14, 5, 5),
(0, "male", 22, 0.125, "no", 1, 16, 3, 5),
(0, "female", 27, 7, "yes", 4, 14, 1, 4),
(0, "female", 32, 15, "yes", 5, 16, 5, 3),
(0, "male", 32, 10, "yes", 4, 18, 5, 4),
(0, "female", 32, 15, "yes", 2, 14, 3, 4),
(0, "female", 22, 1.5, "no", 3, 17, 5, 5),
(0, "male", 27, 4, "yes", 4, 17, 4, 4),
(0, "female", 52, 15, "yes", 5, 14, 1, 5),
(0, "female", 27, 7, "yes", 2, 12, 1, 2),
(0, "female", 27, 7, "yes", 3, 12, 1, 4),
(0, "female", 42, 15, "yes", 2, 14, 1, 4),
(0, "female", 42, 15, "yes", 4, 14, 5, 4),
(0, "male", 27, 7, "yes", 4, 14, 3, 3),
(0, "male", 27, 7, "yes", 2, 20, 6, 2),
(0, "female", 42, 15, "yes", 3, 12, 3, 3),
(0, "male", 27, 4, "yes", 3, 16, 3, 5),
(0, "female", 27, 7, "yes", 3, 14, 1, 4),
(0, "female", 22, 1.5, "no", 2, 14, 4, 5),
(0, "female", 27, 4, "yes", 4, 14, 1, 4),
(0, "female", 22, 4, "no", 4, 14, 5, 5),
(0, "female", 22, 1.5, "no", 2, 16, 4, 5),
(0, "male", 47, 15, "no", 4, 14, 5, 4),
(0, "male", 37, 10, "yes", 2, 18, 6, 2),
(0, "male", 37, 15, "yes", 3, 17, 5, 4),
(0, "female", 27, 4, "yes", 2, 16, 1, 4),
(3, "male", 27, 1.5, "no", 3, 18, 4, 4),
(3, "female", 27, 4, "yes", 3, 17, 1, 5),
(7, "male", 37, 15, "yes", 5, 18, 6, 2),
(12, "female", 32, 10, "yes", 3, 17, 5, 2),
(1, "male", 22, 0.125, "no", 4, 16, 5, 5),
(1, "female", 22, 1.5, "yes", 2, 14, 1, 5),
(12, "male", 37, 15, "yes", 4, 14, 5, 2),
(7, "female", 22, 1.5, "no", 2, 14, 3, 4),
(2, "male", 37, 15, "yes", 2, 18, 6, 4),
(3, "female", 32, 15, "yes", 4, 12, 3, 2),
(1, "female", 37, 15, "yes", 4, 14, 4, 2),
(7, "female", 42, 15, "yes", 3, 17, 1, 4),
(12, "female", 42, 15, "yes", 5, 9, 4, 1),
(12, "male", 37, 10, "yes", 2, 20, 6, 2),
(12, "female", 32, 15, "yes", 3, 14, 1, 2),
(3, "male", 27, 4, "no", 1, 18, 6, 5),
(7, "male", 37, 10, "yes", 2, 18, 7, 3),
(7, "female", 27, 4, "no", 3, 17, 5, 5),
(1, "male", 42, 15, "yes", 4, 16, 5, 5),
(1, "female", 47, 15, "yes", 5, 14, 4, 5),
(7, "female", 27, 4, "yes", 3, 18, 5, 4),
(1, "female", 27, 7, "yes", 5, 14, 1, 4),
(12, "male", 27, 1.5, "yes", 3, 17, 5, 4),
(12, "female", 27, 7, "yes", 4, 14, 6, 2),
(3, "female", 42, 15, "yes", 4, 16, 5, 4),
(7, "female", 27, 10, "yes", 4, 12, 7, 3),
(1, "male", 27, 1.5, "no", 2, 18, 5, 2),
(1, "male", 32, 4, "no", 4, 20, 6, 4),
(1, "female", 27, 7, "yes", 3, 14, 1, 3),
(3, "female", 32, 10, "yes", 4, 14, 1, 4),
(3, "male", 27, 4, "yes", 2, 18, 7, 2),
(1, "female", 17.5, 0.75, "no", 5, 14, 4, 5),
(1, "female", 32, 10, "yes", 4, 18, 1, 5),
(7, "female", 32, 7, "yes", 2, 17, 6, 4),
(7, "male", 37, 15, "yes", 2, 20, 6, 4),
(7, "female", 37, 10, "no", 1, 20, 5, 3),
(12, "female", 32, 10, "yes", 2, 16, 5, 5),
(7, "male", 52, 15, "yes", 2, 20, 6, 4),
(7, "female", 42, 15, "yes", 1, 12, 1, 3),
(1, "male", 52, 15, "yes", 2, 20, 6, 3),
(2, "male", 37, 15, "yes", 3, 18, 6, 5),
(12, "female", 22, 4, "no", 3, 12, 3, 4),
(12, "male", 27, 7, "yes", 1, 18, 6, 2),
(1, "male", 27, 4, "yes", 3, 18, 5, 5),
(12, "male", 47, 15, "yes", 4, 17, 6, 5),
(12, "female", 42, 15, "yes", 4, 12, 1, 1),
(7, "male", 27, 4, "no", 3, 14, 3, 4),
(7, "female", 32, 7, "yes", 4, 18, 4, 5),
(1, "male", 32, 0.417, "yes", 3, 12, 3, 4),
(3, "male", 47, 15, "yes", 5, 16, 5, 4),
(12, "male", 37, 15, "yes", 2, 20, 5, 4),
(7, "male", 22, 4, "yes", 2, 17, 6, 4),
(1, "male", 27, 4, "no", 2, 14, 4, 5),
(7, "female", 52, 15, "yes", 5, 16, 1, 3),
(1, "male", 27, 4, "no", 3, 14, 3, 3),
(1, "female", 27, 10, "yes", 4, 16, 1, 4),
(1, "male", 32, 7, "yes", 3, 14, 7, 4),
(7, "male", 32, 7, "yes", 2, 18, 4, 1),
(3, "male", 22, 1.5, "no", 1, 14, 3, 2),
(7, "male", 22, 4, "yes", 3, 18, 6, 4),
(7, "male", 42, 15, "yes", 4, 20, 6, 4),
(2, "female", 57, 15, "yes", 1, 18, 5, 4),
(7, "female", 32, 4, "yes", 3, 18, 5, 2),
(1, "male", 27, 4, "yes", 1, 16, 4, 4),
(7, "male", 32, 7, "yes", 4, 16, 1, 4),
(2, "male", 57, 15, "yes", 1, 17, 4, 4),
(7, "female", 42, 15, "yes", 4, 14, 5, 2),
(7, "male", 37, 10, "yes", 1, 18, 5, 3),
(3, "male", 42, 15, "yes", 3, 17, 6, 1),
(1, "female", 52, 15, "yes", 3, 14, 4, 4),
(2, "female", 27, 7, "yes", 3, 17, 5, 3),
(12, "male", 32, 7, "yes", 2, 12, 4, 2),
(1, "male", 22, 4, "no", 4, 14, 2, 5),
(3, "male", 27, 7, "yes", 3, 18, 6, 4),
(12, "female", 37, 15, "yes", 1, 18, 5, 5),
(7, "female", 32, 15, "yes", 3, 17, 1, 3),
(7, "female", 27, 7, "no", 2, 17, 5, 5),
(1, "female", 32, 7, "yes", 3, 17, 5, 3),
(1, "male", 32, 1.5, "yes", 2, 14, 2, 4),
(12, "female", 42, 15, "yes", 4, 14, 1, 2),
(7, "male", 32, 10, "yes", 3, 14, 5, 4),
(7, "male", 37, 4, "yes", 1, 20, 6, 3),
(1, "female", 27, 4, "yes", 2, 16, 5, 3),
(12, "female", 42, 15, "yes", 3, 14, 4, 3),
(1, "male", 27, 10, "yes", 5, 20, 6, 5),
(12, "male", 37, 10, "yes", 2, 20, 6, 2),
(12, "female", 27, 7, "yes", 1, 14, 3, 3),
(3, "female", 27, 7, "yes", 4, 12, 1, 2),
(3, "male", 32, 10, "yes", 2, 14, 4, 4),
(12, "female", 17.5, 0.75, "yes", 2, 12, 1, 3),
(12, "female", 32, 15, "yes", 3, 18, 5, 4),
(2, "female", 22, 7, "no", 4, 14, 4, 3),
(1, "male", 32, 7, "yes", 4, 20, 6, 5),
(7, "male", 27, 4, "yes", 2, 18, 6, 2),
(1, "female", 22, 1.5, "yes", 5, 14, 5, 3),
(12, "female", 32, 15, "no", 3, 17, 5, 1),
(12, "female", 42, 15, "yes", 2, 12, 1, 2),
(7, "male", 42, 15, "yes", 3, 20, 5, 4),
(12, "male", 32, 10, "no", 2, 18, 4, 2),
(12, "female", 32, 15, "yes", 3, 9, 1, 1),
(7, "male", 57, 15, "yes", 5, 20, 4, 5),
(12, "male", 47, 15, "yes", 4, 20, 6, 4),
(2, "female", 42, 15, "yes", 2, 17, 6, 3),
(12, "male", 37, 15, "yes", 3, 17, 6, 3),
(12, "male", 37, 15, "yes", 5, 17, 5, 2),
(7, "male", 27, 10, "yes", 2, 20, 6, 4),
(2, "male", 37, 15, "yes", 2, 16, 5, 4),
(12, "female", 32, 15, "yes", 1, 14, 5, 2),
(7, "male", 32, 10, "yes", 3, 17, 6, 3),
(2, "male", 37, 15, "yes", 4, 18, 5, 1),
(7, "female", 27, 1.5, "no", 2, 17, 5, 5),
(3, "female", 47, 15, "yes", 2, 17, 5, 2),
(12, "male", 37, 15, "yes", 2, 17, 5, 4),
(12, "female", 27, 4, "no", 2, 14, 5, 5),
(2, "female", 27, 10, "yes", 4, 14, 1, 5),
(1, "female", 22, 4, "yes", 3, 16, 1, 3),
(12, "male", 52, 7, "no", 4, 16, 5, 5),
(2, "female", 27, 4, "yes", 1, 16, 3, 5),
(7, "female", 37, 15, "yes", 2, 17, 6, 4),
(2, "female", 27, 4, "no", 1, 17, 3, 1),
(12, "female", 17.5, 0.75, "yes", 2, 12, 3, 5),
(7, "female", 32, 15, "yes", 5, 18, 5, 4),
(7, "female", 22, 4, "no", 1, 16, 3, 5),
(2, "male", 32, 4, "yes", 4, 18, 6, 4),
(1, "female", 22, 1.5, "yes", 3, 18, 5, 2),
(3, "female", 42, 15, "yes", 2, 17, 5, 4),
(1, "male", 32, 7, "yes", 4, 16, 4, 4),
(12, "male", 37, 15, "no", 3, 14, 6, 2),
(1, "male", 42, 15, "yes", 3, 16, 6, 3),
(1, "male", 27, 4, "yes", 1, 18, 5, 4),
(2, "male", 37, 15, "yes", 4, 20, 7, 3),
(7, "male", 37, 15, "yes", 3, 20, 6, 4),
(3, "male", 22, 1.5, "no", 2, 12, 3, 3),
(3, "male", 32, 4, "yes", 3, 20, 6, 2),
(2, "male", 32, 15, "yes", 5, 20, 6, 5),
(12, "female", 52, 15, "yes", 1, 18, 5, 5),
(12, "male", 47, 15, "no", 1, 18, 6, 5),
(3, "female", 32, 15, "yes", 4, 16, 4, 4),
(7, "female", 32, 15, "yes", 3, 14, 3, 2),
(7, "female", 27, 7, "yes", 4, 16, 1, 2),
(12, "male", 42, 15, "yes", 3, 18, 6, 2),
(7, "female", 42, 15, "yes", 2, 14, 3, 2),
(12, "male", 27, 7, "yes", 2, 17, 5, 4),
(3, "male", 32, 10, "yes", 4, 14, 4, 3),
(7, "male", 47, 15, "yes", 3, 16, 4, 2),
(1, "male", 22, 1.5, "yes", 1, 12, 2, 5),
(7, "female", 32, 10, "yes", 2, 18, 5, 4),
(2, "male", 32, 10, "yes", 2, 17, 6, 5),
(2, "male", 22, 7, "yes", 3, 18, 6, 2),
(1, "female", 32, 15, "yes", 3, 14, 1, 5))
val colArray1: Array[String] = Array("affairs", "gender", "age", "yearsmarried", "children", "religiousness", "education", "occupation", "rating")
val data = dataList.toDF(colArray1:_*)
建立多層感知器分類器MLPC模型
data.createOrReplaceTempView("data")
// 字元型別轉換成數值
val labelWhere = "case when affairs=0 then 0 else cast(1 as double) end as label"
val genderWhere = "case when gender='female' then 0 else cast(1 as double) end as gender"
val childrenWhere = "case when children='no' then 0 else cast(1 as double) end as children"
val dataLabelDF = spark.sql(s"select $labelWhere, $genderWhere,age,yearsmarried,$childrenWhere,religiousness,education,occupation,rating from data")
val featuresArray = Array("gender", "age", "yearsmarried", "children", "religiousness", "education", "occupation", "rating")
// 欄位轉換成特徵向量
val assembler = new VectorAssembler().setInputCols(featuresArray).setOutputCol("features")
val vecDF: DataFrame = assembler.transform(dataLabelDF)
vecDF.show(10, truncate = false)
// 分割資料
val splits = vecDF.randomSplit(Array(0.6, 0.4), seed = 1234L)
val trainDF = splits(0)
val testDF = splits(1)
// 隱藏層結點數=2n+1，n為輸入結點數
// 指定神經網路的圖層：輸入層8個節點(即8個特徵)；兩個隱藏層，隱藏結點數分別為9和8；輸出層2個結點(即二分類)
val layers = Array[Int](8, 9, 8, 2)
// 建立多層感知器分類器MLPC模型
// 傳統神經網路通常，層數<=5，隱藏層數<=3
// layers 指定神經網路的圖層
// MaxIter 最大迭代次數
// stepSize 每次優化的迭代步長，僅適用於solver="gd"
// blockSize 用於在矩陣中堆疊輸入資料的塊大小以加速計算。 資料在分割槽內堆疊。 如果塊大小大於分割槽中的剩餘資料，則將其調整為該資料的大小。 建議大小介於10到1000之間。預設值：128
// initialWeights 模型的初始權重
// solver 演算法優化。 支援的選項：“gd”（minibatch梯度下降）或“l-bfgs”。 預設值：“l-bfgs”
val trainer = new MultilayerPerceptronClassifier().setFeaturesCol("features").setLabelCol("label").setLayers(layers)
//.setMaxIter(100).setTol(1E-4).setSeed(1234L)
//.setBlockSize(128).setSolver("l-bfgs")
//.setInitialWeights(Vector).setStepSize(0.03)
// 訓練模型
val model = trainer.fit(trainDF)
// 測試
val result = model.transform(testDF)
val predictionLabels = result.select("prediction", "label")
// 計算精度
val evaluator = new MulticlassClassificationEvaluator().setPredictionCol("prediction").setLabelCol("label").setMetricName("accuracy")
println("Accuracy: " + evaluator.evaluate(predictionLabels))
程式碼執行結果
vecDF.show(10, truncate = false)
+-----+------+----+------------+--------+-------------+---------+----------+------+------------------------------------+
|label|gender|age |yearsmarried|children|religiousness|education|occupation|rating|features |
+-----+------+----+------------+--------+-------------+---------+----------+------+------------------------------------+
|0.0 |1.0 |37.0|10.0 |0.0 |3.0 |18.0 |7.0 |4.0 |[1.0,37.0,10.0,0.0,3.0,18.0,7.0,4.0]|
|0.0 |0.0 |27.0|4.0 |0.0 |4.0 |14.0 |6.0 |4.0 |[0.0,27.0,4.0,0.0,4.0,14.0,6.0,4.0] |
|0.0 |0.0 |32.0|15.0 |1.0 |1.0 |12.0 |1.0 |4.0 |[0.0,32.0,15.0,1.0,1.0,12.0,1.0,4.0]|
|0.0 |1.0 |57.0|15.0 |1.0 |5.0 |18.0 |6.0 |5.0 |[1.0,57.0,15.0,1.0,5.0,18.0,6.0,5.0]|
|0.0 |1.0 |22.0|0.75 |0.0 |2.0 |17.0 |6.0 |3.0 |[1.0,22.0,0.75,0.0,2.0,17.0,6.0,3.0]|
|0.0 |0.0 |32.0|1.5 |0.0 |2.0 |17.0 |5.0 |5.0 |[0.0,32.0,1.5,0.0,2.0,17.0,5.0,5.0] |
|0.0 |0.0 |22.0|0.75 |0.0 |2.0 |12.0 |1.0 |3.0 |[0.0,22.0,0.75,0.0,2.0,12.0,1.0,3.0]|
|0.0 |1.0 |57.0|15.0 |1.0 |2.0 |14.0 |4.0 |4.0 |[1.0,57.0,15.0,1.0,2.0,14.0,4.0,4.0]|
|0.0 |0.0 |32.0|15.0 |1.0 |4.0 |16.0 |1.0 |2.0 |[0.0,32.0,15.0,1.0,4.0,16.0,1.0,2.0]|
|0.0 |1.0 |22.0|1.5 |0.0 |4.0 |14.0 |4.0 |5.0 |[1.0,22.0,1.5,0.0,4.0,14.0,4.0,5.0] |
+-----+------+----+------------+--------+-------------+---------+----------+------+------------------------------------+
only showing top 10 rows
// 分割資料
val splits = vecDF.randomSplit(Array(0.6, 0.4), seed = 1234L)
splits: Array[org.apache.spark.sql.Dataset[org.apache.spark.sql.Row]] = Array([label: double, gender: double ... 8 more fields], [label: double, gender: double ... 8 more fields])
val trainDF = splits(0)
trainDF: org.apache.spark.sql.Dataset[org.apache.spark.sql.Row] = [label: double, gender: double ... 8 more fields]
val testDF = splits(1)
testDF: org.apache.spark.sql.Dataset[org.apache.spark.sql.Row] = [label: double, gender: double ... 8 more fields]
// 隱藏層結點數=2n+1，n為輸入結點數
// 指定神經網路的圖層：輸入層8個節點(即8個特徵)；兩個隱藏層，隱藏結點數分別為9和8；輸出層2個結點(即二分類)
val layers = Array[Int](8, 9, 8, 2)
layers: Array[Int] = Array(8, 9, 8, 2)
// 建立多層感知器分類器MLPC模型
// 傳統神經網路通常，層數<=5，隱藏層數<=3
// layers 指定神經網路的圖層
// MaxIter 最大迭代次數
// stepSize 每次優化的迭代步長，僅適用於solver="gd"
// blockSize 用於在矩陣中堆疊輸入資料的塊大小以加速計算。 資料在分割槽內堆疊。 如果塊大小大於分割槽中的剩餘資料，則將其調整為該資料的大小。 建議大小介於10到1000之間。預設值：128
// initialWeights 模型的初始權重
// solver 演算法優化。 支援的選項：“gd”（minibatch梯度下降）或“l-bfgs”。 預設值：“l-bfgs”
val trainer = new MultilayerPerceptronClassifier().setFeaturesCol("features").setLabelCol("label").setLayers(layers)
//.setMaxIter(100).setTol(1E-4).setSeed(1234L)
//.setBlockSize(128).setSolver("l-bfgs")
//.setInitialWeights(Vector).setStepSize(0.03)
// 訓練模型
val model = trainer.fit(trainDF)
// 測試
val result = model.transform(testDF)
result: org.apache.spark.sql.DataFrame = [label: double, gender: double ... 9 more fields]
val predictionLabels = result.select("prediction", "label")
predictionLabels: org.apache.spark.sql.DataFrame = [prediction: double, label: double]
// 計算精度
val evaluator = new MulticlassClassificationEvaluator().setPredictionCol("prediction").setLabelCol("label").setMetricName("accuracy")
println("Accuracy: " + evaluator.evaluate(predictionLabels))
Accuracy: 0.7229437229437229
model.extractParamMap()
res7: org.apache.spark.ml.param.ParamMap =
{
mlpc_27ec9285cc65-featuresCol: features,
mlpc_27ec9285cc65-labelCol: label,
mlpc_27ec9285cc65-predictionCol: prediction
}
model.layers
res8: Array[Int] = Array(8, 9, 8, 2)
model.numFeatures
res9: Int = 8
model.weights
res10: org.apache.spark.ml.linalg.Vector = [0.6687947649918738,0.19748283690078897,-50.45651427272606,-45.341284166506405,-0.6919873953519974,-0.26471957984972305,0.23736474165465096,0.436660
4816264614,-0.710245071766165,-0.5988483140879621,-1.7544682663368922,15.823927446929437,-22.032609932367354,0.22889325554600204,-0.6427955112694319,-0.3723715157477082,0.7674184433539617,-0.3716890493835326,0.36611178436831204,-0.044164649859209495,-21.138120634483425,-41.97482820667807,0.8325520012775748,0.04477409210962596,-0.4552441455426475,0.0350190939749595,-0.06185632639036518,0.5480105570145155,0.44454066597818026,3.0222205130311064,-19.289596176945746,-0.3104490505088977,-0.77751632210236,-0.05028469739291188,-0.4138707133706901,0.6942105256488522,-0.00808567540478167,0.24805364902280116,2...
result.show(20,false)
+-----+------+----+------------+--------+-------------+---------+----------+------+-------------------------------------+----------+
|label|gender|age |yearsmarried|children|religiousness|education|occupation|rating|features |prediction|
+-----+------+----+------------+--------+-------------+---------+----------+------+-------------------------------------+----------+
|0.0 |0.0 |22.0|0.125 |0.0 |2.0 |12.0 |5.0 |5.0 |[0.0,22.0,0.125,0.0,2.0,12.0,5.0,5.0]|0.0 |
|0.0 |0.0 |22.0|0.125 |0.0 |2.0 |14.0 |4.0 |5.0 |[0.0,22.0,0.125,0.0,2.0,14.0,4.0,5.0]|0.0 |
|0.0 |0.0 |22.0|0.125 |0.0 |2.0 |16.0 |6.0 |3.0 |[0.0,22.0,0.125,0.0,2.0,16.0,6.0,3.0]|0.0 |
|0.0 |0.0 |22.0|0.417 |0.0 |5.0 |14.0 |1.0 |4.0 |[0.0,22.0,0.417,0.0,5.0,14.0,1.0,4.0]|0.0 |
|0.0 |0.0 |22.0|0.417 |1.0 |3.0 |14.0 |3.0 |5.0 |[0.0,22.0,0.417,1.0,3.0,14.0,3.0,5.0]|0.0 |
|0.0 |0.0 |22.0|0.75 |0.0 |2.0 |16.0 |3.0 |4.0 |[0.0,22.0,0.75,0.0,2.0,16.0,3.0,4.0] |0.0 |
|0.0 |0.0 |22.0|0.75 |0.0 |3.0 |16.0 |1.0 |5.0 |[0.0,22.0,0.75,0.0,3.0,16.0,1.0,5.0] |0.0 |
|0.0 |0.0 |22.0|1.5 |0.0 |2.0 |16.0 |5.0 |5.0 |[0.0,22.0,1.5,0.0,2.0,16.0,5.0,5.0] |0.0 |
|0.0 |0.0 |22.0|1.5 |0.0 |2.0 |17.0 |5.0 |4.0 |[0.0,22.0,1.5,0.0,2.0,17.0,5.0,4.0] |0.0 |
|0.0 |0.0 |22.0|1.5 |0.0 |2.0 |18.0 |5.0 |5.0 |[0.0,22.0,1.5,0.0,2.0,18.0,5.0,5.0] |0.0 |
|0.0 |0.0 |22.0|1.5 |0.0 |3.0 |16.0 |5.0 |5.0 |[0.0,22.0,1.5,0.0,3.0,16.0,5.0,5.0] |0.0 |
|0.0 |0.0 |22.0|1.5 |0.0 |3.0 |16.0 |6.0 |5.0 |[0.0,22.0,1.5,0.0,3.0,16.0,6.0,5.0] |0.0 |
|0.0 |0.0 |22.0|1.5 |0.0 |4.0 |16.0 |5.0 |3.0 |[0.0,22.0,1.5,0.0,4.0,16.0,5.0,3.0] |0.0 |
|0.0 |0.0 |22.0|1.5 |1.0 |3.0 |12.0 |1.0 |3.0 |[0.0,22.0,1.5,1.0,3.0,12.0,1.0,3.0] |0.0 |
|0.0 |0.0 |22.0|1.5 |1.0 |4.0 |12.0 |1.0 |5.0 |[0.0,22.0,1.5,1.0,4.0,12.0,1.0,5.0] |0.0 |
|0.0 |0.0 |22.0|4.0 |1.0 |3.0 |9.0 |1.0 |4.0 |[0.0,22.0,4.0,1.0,3.0,9.0,1.0,4.0] |0.0 |
|0.0 |0.0 |22.0|4.0 |1.0 |3.0 |14.0 |1.0 |5.0 |[0.0,22.0,4.0,1.0,3.0,14.0,1.0,5.0] |0.0 |
|0.0 |0.0 |22.0|7.0 |1.0 |2.0 |14.0 |5.0 |2.0 |[0.0,22.0,7.0,1.0,2.0,14.0,5.0,2.0] |0.0 |
|0.0 |0.0 |27.0|1.5 |0.0 |2.0 |18.0 |6.0 |5.0 |[0.0,27.0,1.5,0.0,2.0,18.0,6.0,5.0] |0.0 |
|0.0 |0.0 |27.0|4.0 |0.0 |3.0 |17.0 |5.0 |5.0 |[0.0,27.0,4.0,0.0,3.0,17.0,5.0,5.0] |0.0 |
+-----+------+----+------------+--------+-------------+---------+----------+------+-------------------------------------+----------+
only showing top 20 rows