From 9c7f34af37ef328149c1d66b4689d80a1589e1cc Mon Sep 17 00:00:00 2001 From: Sean Owen Date: Tue, 12 Jan 2016 12:13:32 +0000 Subject: [SPARK-5273][MLLIB][DOCS] Improve documentation examples for LinearRegression Use a much smaller step size in LinearRegressionWithSGD MLlib examples to achieve a reasonable RMSE. Our training folks hit this exact same issue when concocting an example and had the same solution. Author: Sean Owen Closes #10675 from srowen/SPARK-5273. --- docs/mllib-linear-methods.md | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'docs') diff --git a/docs/mllib-linear-methods.md b/docs/mllib-linear-methods.md index 20b35612ca..aac8f7560a 100644 --- a/docs/mllib-linear-methods.md +++ b/docs/mllib-linear-methods.md @@ -590,7 +590,8 @@ val parsedData = data.map { line => // Building the model val numIterations = 100 -val model = LinearRegressionWithSGD.train(parsedData, numIterations) +val stepSize = 0.00000001 +val model = LinearRegressionWithSGD.train(parsedData, numIterations, stepSize) // Evaluate model on training examples and compute training error val valuesAndPreds = parsedData.map { point => @@ -655,8 +656,9 @@ public class LinearRegression { // Building the model int numIterations = 100; + double stepSize = 0.00000001; final LinearRegressionModel model = - LinearRegressionWithSGD.train(JavaRDD.toRDD(parsedData), numIterations); + LinearRegressionWithSGD.train(JavaRDD.toRDD(parsedData), numIterations, stepSize); // Evaluate model on training examples and compute training error JavaRDD> valuesAndPreds = parsedData.map( @@ -706,7 +708,7 @@ data = sc.textFile("data/mllib/ridge-data/lpsa.data") parsedData = data.map(parsePoint) # Build the model -model = LinearRegressionWithSGD.train(parsedData) +model = LinearRegressionWithSGD.train(parsedData, iterations=100, step=0.00000001) # Evaluate the model on training data valuesAndPreds = parsedData.map(lambda p: (p.label, model.predict(p.features))) -- cgit v1.2.3