Re: Storing an action result in HDFS
Hi Ravi, Welcome, you probably want RDD.saveAsTextFile(“hdfs:///my_file”) Chris On Jun 22, 2015, at 5:28 PM, ravi tella ddpis...@gmail.com wrote: Hello All, I am new to Spark. I have a very basic question.How do I write the output of an action on a RDD to HDFS? Thanks in advance for the help. Cheers, Ravi - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: Storing an action result in HDFS
Hi Ravi, For this case, you could simply do sc.parallelize([rdd.first()]).saveAsTextFile(“hdfs:///my_file”) using pyspark or sc.parallelize(Array(rdd.first())).saveAsTextFile(“hdfs:///my_file”) using Scala Chris On Jun 22, 2015, at 5:53 PM, ddpis...@gmail.com wrote: Hi Chris, Thanks for the quick reply and the welcome. I am trying to read a file from hdfs and then writing back just the first line to hdfs. I calling first() on the RDD to get the first line. Sent from my iPhone On Jun 22, 2015, at 7:42 PM, Chris Gore cdg...@cdgore.com wrote: Hi Ravi, Welcome, you probably want RDD.saveAsTextFile(“hdfs:///my_file”) Chris On Jun 22, 2015, at 5:28 PM, ravi tella ddpis...@gmail.com wrote: Hello All, I am new to Spark. I have a very basic question.How do I write the output of an action on a RDD to HDFS? Thanks in advance for the help. Cheers, Ravi - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: Compare LogisticRegression results using Mllib with those using other libraries (e.g. statsmodel)
I tried running this data set as described with my own implementation of L2 regularized logistic regression using LBFGS to compare: https://github.com/cdgore/fitbox https://github.com/cdgore/fitbox Intercept: -0.886745823033 Weights (['gre', 'gpa', 'rank']):[ 0.28862268 0.19402388 -0.36637964] Area under ROC: 0.724056603774 The difference could be from the feature preprocessing as mentioned. I normalized the features around 0: binary_train_normalized = (binary_train - binary_train.mean()) / binary_train.std() binary_test_normalized = (binary_test - binary_train.mean()) / binary_train.std() On a data set this small, the difference in models could also be the result of how the training/test sets were split. Have you tried running k-folds cross validation on a larger data set? Chris On May 20, 2015, at 6:15 PM, DB Tsai d...@netflix.com.INVALID wrote: Hi Xin, If you take a look at the model you trained, the intercept from Spark is significantly smaller than StatsModel, and the intercept represents a prior on categories in LOR which causes the low accuracy in Spark implementation. In LogisticRegressionWithLBFGS, the intercept is regularized due to the implementation of Updater, and the intercept should not be regularized. In the new pipleline APIs, a LOR with elasticNet is implemented, and the intercept is properly handled. https://github.com/apache/spark/blob/master/mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala As you can see the tests, https://github.com/apache/spark/blob/master/mllib/src/test/scala/org/apache/spark/ml/classification/LogisticRegressionSuite.scala the result is exactly the same as R now. BTW, in both version, the feature scalings are done before training, and we train the model in scaled space but transform the model weights back to original space. The only difference is in the mllib version, LogisticRegressionWithLBFGS regularizes the intercept while in the ml version, the intercept is excluded from regularization. As a result, if lambda is zero, the model should be the same. On Wed, May 20, 2015 at 3:42 PM, Xin Liu liuxin...@gmail.com wrote: Hi, I have tried a few models in Mllib to train a LogisticRegression model. However, I consistently get much better results using other libraries such as statsmodel (which gives similar results as R) in terms of AUC. For illustration purpose, I used a small data (I have tried much bigger data) http://www.ats.ucla.edu/stat/data/binary.csv in http://www.ats.ucla.edu/stat/r/dae/logit.htm Here is the snippet of my usage of LogisticRegressionWithLBFGS. val algorithm = new LogisticRegressionWithLBFGS algorithm.setIntercept(true) algorithm.optimizer .setNumIterations(100) .setRegParam(0.01) .setConvergenceTol(1e-5) val model = algorithm.run(training) model.clearThreshold() val scoreAndLabels = test.map { point = val score = model.predict(point.features) (score, point.label) } val metrics = new BinaryClassificationMetrics(scoreAndLabels) val auROC = metrics.areaUnderROC() I did a (0.6, 0.4) split for training/test. The response is admit and features are GRE score, GPA, and college Rank. Spark: Weights (GRE, GPA, Rank): [0.0011576276331509304,0.048544858567336854,-0.394202150286076] Intercept: -0.6488972641282202 Area under ROC: 0.6294070512820512 StatsModel: Weights [0.0018, 0.7220, -0.3148] Intercept: -3.5913 Area under ROC: 0.69 The weights from statsmodel seems more reasonable if you consider for a one unit increase in gpa, the log odds of being admitted to graduate school increases by 0.72 in statsmodel than 0.04 in Spark. I have seen much bigger difference with other data. So my question is has anyone compared the results with other libraries and is anything wrong with my code to invoke LogisticRegressionWithLBFGS? As the real data I am processing is pretty big and really want to use Spark to get this to work. Please let me know if you have similar experience and how you resolve it. Thanks, Xin - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: Can Spark benefit from Hive-like partitions?
Good to hear there will be partitioning support. I’ve had some success loading partitioned data specified with Unix glowing format. i.e.: sc.textFile(s3:/bucket/directory/dt=2014-11-{2[4-9],30}T00-00-00”) would load dates 2014-11-24 through 2014-11-30. Not the most ideal solution, but it seems to work for loading data from a range. Best, Chris On Jan 26, 2015, at 10:55 AM, Cheng Lian lian.cs@gmail.com wrote: Currently no if you don't want to use Spark SQL's HiveContext. But we're working on adding partitioning support to the external data sources API, with which you can create, for example, partitioned Parquet tables without using Hive. Cheng On 1/26/15 8:47 AM, Danny Yates wrote: Thanks Michael. I'm not actually using Hive at the moment - in fact, I'm trying to avoid it if I can. I'm just wondering whether Spark has anything similar I can leverage? Thanks - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: MLLib sparse vector
Hi Sameer, MLLib uses Breeze’s vector format under the hood. You can use that. http://www.scalanlp.org/api/breeze/index.html#breeze.linalg.SparseVector For example: import breeze.linalg.{DenseVector = BDV, SparseVector = BSV, Vector = BV} val numClasses = classes.distinct.count.toInt val userWithClassesAsSparseVector = rows.map(x = (x.userID, new BSV[Double](x.classIDs.sortWith(_ _), Seq.fill(x.classIDs.length)(1.0).toArray, numClasses).asInstanceOf[BV[Double]])) Chris On Sep 15, 2014, at 11:28 AM, Sameer Tilak ssti...@live.com wrote: Hi All, I have transformed the data into following format: First column is user id, and then all the other columns are class ids. For a user only class ids that appear in this row have value 1 and others are 0. I need to crease a sparse vector from this. Does the API for creating a sparse vector that can directly support this format? User idProduct class ids 2622572 145447 162013421 28565 285556 293 455367261 130 3646167118806 183576 328651715 57671 57476
Re: MLLib sparse vector
Probably worth noting that the factory methods in mllib create an object of type org.apache.spark.mllib.linalg.Vector which stores data in a similar format as Breeze vectors Chris On Sep 15, 2014, at 3:24 PM, Xiangrui Meng men...@gmail.com wrote: Or you can use the factory method `Vectors.sparse`: val sv = Vectors.sparse(numProducts, productIds.map(x = (x, 1.0))) where numProducts should be the largest product id plus one. Best, Xiangrui On Mon, Sep 15, 2014 at 12:46 PM, Chris Gore cdg...@cdgore.com wrote: Hi Sameer, MLLib uses Breeze’s vector format under the hood. You can use that. http://www.scalanlp.org/api/breeze/index.html#breeze.linalg.SparseVector For example: import breeze.linalg.{DenseVector = BDV, SparseVector = BSV, Vector = BV} val numClasses = classes.distinct.count.toInt val userWithClassesAsSparseVector = rows.map(x = (x.userID, new BSV[Double](x.classIDs.sortWith(_ _), Seq.fill(x.classIDs.length)(1.0).toArray, numClasses).asInstanceOf[BV[Double]])) Chris On Sep 15, 2014, at 11:28 AM, Sameer Tilak ssti...@live.com wrote: Hi All, I have transformed the data into following format: First column is user id, and then all the other columns are class ids. For a user only class ids that appear in this row have value 1 and others are 0. I need to crease a sparse vector from this. Does the API for creating a sparse vector that can directly support this format? User idProduct class ids 2622572 145447 1620 13421 28565 285556 293 4553 67261 130 3646 1671 18806 183576 3286 51715 57671 57476 - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: Accessing neighboring elements in an RDD
There is support for Spark in ElasticSearch’s Hadoop integration package. http://www.elasticsearch.org/guide/en/elasticsearch/hadoop/current/spark.html Maybe you could split and insert all of your documents from Spark and then query for “MoreLikeThis” on the ElasticSearch index. I haven’t tried it, but maybe someone else has more experience using Spark with ElasticSearch. At some point, maybe there could be an information retrieval package for Spark with locality sensitive hashing and other similar functions. On Sep 3, 2014, at 10:40 AM, Victor Tso-Guillen v...@paxata.com wrote: Interestingly, there was an almost identical question posed on Aug 22 by cjwang. Here's the link to the archive: http://apache-spark-user-list.1001560.n3.nabble.com/Finding-previous-and-next-element-in-a-sorted-RDD-td12621.html#a12664 On Wed, Sep 3, 2014 at 10:33 AM, Daniel, Ronald (ELS-SDG) r.dan...@elsevier.com wrote: Hi all, Assume I have read the lines of a text file into an RDD: textFile = sc.textFile(SomeArticle.txt) Also assume that the sentence breaks in SomeArticle.txt were done by machine and have some errors, such as the break at Fig. in the sample text below. Index Text N...as shown in Fig. N+1 1. N+2 The figure shows... What I want is an RDD with: N ... as shown in Fig. 1. N+1 The figure shows... Is there some way a filter() can look at neighboring elements in an RDD? That way I could look, in parallel, at neighboring elements in an RDD and come up with a new RDD that may have a different number of elements. Or do I just have to sequentially iterate through the RDD? Thanks, Ron
Re: Error: No space left on device
Hi Chris, I've encountered this error when running Spark’s ALS methods too. In my case, it was because I set spark.local.dir improperly, and every time there was a shuffle, it would spill many GB of data onto the local drive. What fixed it was setting it to use the /mnt directory, where a network drive is mounted. For example, setting an environmental variable: export SPACE=$(mount | grep mnt | awk '{print $3/spark/}' | xargs | sed 's/ /,/g’) Then adding -Dspark.local.dir=$SPACE or simply -Dspark.local.dir=/mnt/spark/,/mnt2/spark/ when you run your driver application Chris On Jul 15, 2014, at 11:39 PM, Xiangrui Meng men...@gmail.com wrote: Check the number of inodes (df -i). The assembly build may create many small files. -Xiangrui On Tue, Jul 15, 2014 at 11:35 PM, Chris DuBois chris.dub...@gmail.com wrote: Hi all, I am encountering the following error: INFO scheduler.TaskSetManager: Loss was due to java.io.IOException: No space left on device [duplicate 4] For each slave, df -h looks roughtly like this, which makes the above error surprising. FilesystemSize Used Avail Use% Mounted on /dev/xvda17.9G 4.4G 3.5G 57% / tmpfs 7.4G 4.0K 7.4G 1% /dev/shm /dev/xvdb 37G 3.3G 32G 10% /mnt /dev/xvdf 37G 2.0G 34G 6% /mnt2 /dev/xvdv 500G 33M 500G 1% /vol I'm on an EC2 cluster (c3.xlarge + 5 x m3) that I launched using the spark-ec2 scripts and a clone of spark from today. The job I am running closely resembles the collaborative filtering example. This issue happens with the 1M version as well as the 10 million rating version of the MovieLens dataset. I have seen previous questions, but they haven't helped yet. For example, I tried setting the Spark tmp directory to the EBS volume at /vol/, both by editing the spark conf file (and copy-dir'ing it to the slaves) as well as through the SparkConf. Yet I still get the above error. Here is my current Spark config below. Note that I'm launching via ~/spark/bin/spark-submit. conf = SparkConf() conf.setAppName(RecommendALS).set(spark.local.dir, /vol/).set(spark.executor.memory, 7g).set(spark.akka.frameSize, 100).setExecutorEnv(SPARK_JAVA_OPTS, -Dspark.akka.frameSize=100) sc = SparkContext(conf=conf) Thanks for any advice, Chris
Re: Calling Spark enthusiasts in NYC
We'd love to see a Spark user group in Los Angeles and connect with others working with it here. Ping me if you're in the LA area and use Spark at your company ( ch...@retentionscience.com ). Chris Retention Science call: 734.272.3099 visit: Site | like: Facebook | follow: Twitter On Mar 31, 2014, at 10:42 AM, Anurag Dodeja anu...@anuragdodeja.com wrote: How about Chicago? On Mon, Mar 31, 2014 at 12:38 PM, Nan Zhu zhunanmcg...@gmail.com wrote: Montreal or Toronto? On Mon, Mar 31, 2014 at 1:36 PM, Martin Goodson mar...@skimlinks.com wrote: How about London? -- Martin Goodson | VP Data Science (0)20 3397 1240 image.png On Mon, Mar 31, 2014 at 6:28 PM, Andy Konwinski andykonwin...@gmail.com wrote: Hi folks, We have seen a lot of community growth outside of the Bay Area and we are looking to help spur even more! For starters, the organizers of the Spark meetups here in the Bay Area want to help anybody that is interested in setting up a meetup in a new city. Some amazing Spark champions have stepped forward in Seattle, Vancouver, Boulder/Denver, and a few other areas already. Right now, we are looking to connect with you Spark enthusiasts in NYC about helping to run an inaugural Spark Meetup in your area. You can reply to me directly if you are interested and I can tell you about all of the resources we have to offer (speakers from the core community, a budget for food, help scheduling, etc.), and let's make this happen! Andy