Spark test error

[Yanwei Wayne Zhang]

I tried to run the tests in 'GeneralizedLinearRegressionSuite', and all tests 
passed except for test("read/write") which yielded the following error message. 
Any suggestion on why this happened and how to fix it? Thanks. BTW, I ran the 
test in IntelliJ.


The default jsonEncode only supports string and vector. 
org.apache.spark.ml.param.Param must override jsonEncode for java.lang.Double.
scala.NotImplementedError: The default jsonEncode only supports string and 
vector. org.apache.spark.ml.param.Param must override jsonEncode for 
java.lang.Double.
at org.apache.spark.ml.param.Param.jsonEncode(params.scala:98)
at 
org.apache.spark.ml.util.DefaultParamsWriter$$anonfun$1$$anonfun$2.apply(ReadWrite.scala:293)
at 
org.apache.spark.ml.util.DefaultParamsWriter$$anonfun$1$$anonfun$2.apply(ReadWrite.scala:292)


Regards,
Wayne

Re: Invert large matrix

[Yanwei Wayne Zhang]

Thanks for the advice. I figured out a way to solve this problem by avoiding 
the matrix representation.


Wayne



From: Sean Owen <so...@cloudera.com>
Sent: Thursday, December 29, 2016 1:52 PM
To: Yanwei Wayne Zhang; user
Subject: Re: Invert large matrix

I think the best advice is: don't do that. If you're trying to solve a linear 
system, solve the linear system without explicitly constructing a matrix 
inverse. Is that what you mean?

On Thu, Dec 29, 2016 at 2:22 AM Yanwei Wayne Zhang 
<actuary_zh...@hotmail.com<mailto:actuary_zh...@hotmail.com>> wrote:

Hi all,


I have a matrix X stored as RDD[SparseVector] that is high dimensional, say 800 
million rows and 2 million columns, and more 95% of the entries are zero.

Is there a way to invert (X'X + eye) efficiently, where X' is the transpose of 
X and eye is the identity matrix? I am thinking of using RowMatrix but not sure 
if it is feasible.

Any suggestion is highly appreciated.


Thanks.


Wayne

Invert large matrix

[Yanwei Wayne Zhang]

Hi all,


I have a matrix X stored as RDD[SparseVector] that is high dimensional, say 800 
million rows and 2 million columns, and more 95% of the entries are zero.

Is there a way to invert (X'X + eye) efficiently, where X' is the transpose of 
X and eye is the identity matrix? I am thinking of using RowMatrix but not sure 
if it is feasible.

Any suggestion is highly appreciated.


Thanks.


Wayne