## What changes were proposed in this pull request?
SparkR ```spark.getSparkFiles``` fails when it was called on executors, see details at [SPARK-19925](https://issues.apache.org/jira/browse/SPARK-19925).
## How was this patch tested?
Add unit tests, and verify this fix at standalone and yarn cluster.
Author: Yanbo Liang <ybliang8@gmail.com>
Closes#17274 from yanboliang/spark-19925.
## What changes were proposed in this pull request?
When SparkR is installed as a R package there might not be any java runtime.
If it is not there SparkR's `sparkR.session()` will block waiting for the connection timeout, hanging the R IDE/shell, without any notification or message.
## How was this patch tested?
manually
- [x] need to test on Windows
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16596 from felixcheung/rcheckjava.
## What changes were proposed in this pull request?
Currently JSON and CSV have exactly the same logic about handling bad records, this PR tries to abstract it and put it in a upper level to reduce code duplication.
The overall idea is, we make the JSON and CSV parser to throw a BadRecordException, then the upper level, FailureSafeParser, handles bad records according to the parse mode.
Behavior changes:
1. with PERMISSIVE mode, if the number of tokens doesn't match the schema, previously CSV parser will treat it as a legal record and parse as many tokens as possible. After this PR, we treat it as an illegal record, and put the raw record string in a special column, but we still parse as many tokens as possible.
2. all logging is removed as they are not very useful in practice.
## How was this patch tested?
existing tests
Author: Wenchen Fan <wenchen@databricks.com>
Author: hyukjinkwon <gurwls223@gmail.com>
Author: Wenchen Fan <cloud0fan@gmail.com>
Closes#17315 from cloud-fan/bad-record2.
## What changes were proposed in this pull request?
Add checkpoint, setCheckpointDir API to R
## How was this patch tested?
unit tests, manual tests
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#17351 from felixcheung/rdfcheckpoint.
## What changes were proposed in this pull request?
This PR proposes to support an array of struct type in `to_json` as below:
```scala
import org.apache.spark.sql.functions._
val df = Seq(Tuple1(Tuple1(1) :: Nil)).toDF("a")
df.select(to_json($"a").as("json")).show()
```
```
+----------+
| json|
+----------+
|[{"_1":1}]|
+----------+
```
Currently, it throws an exception as below (a newline manually inserted for readability):
```
org.apache.spark.sql.AnalysisException: cannot resolve 'structtojson(`array`)' due to data type
mismatch: structtojson requires that the expression is a struct expression.;;
```
This allows the roundtrip with `from_json` as below:
```scala
import org.apache.spark.sql.functions._
import org.apache.spark.sql.types._
val schema = ArrayType(StructType(StructField("a", IntegerType) :: Nil))
val df = Seq("""[{"a":1}, {"a":2}]""").toDF("json").select(from_json($"json", schema).as("array"))
df.show()
// Read back.
df.select(to_json($"array").as("json")).show()
```
```
+----------+
| array|
+----------+
|[[1], [2]]|
+----------+
+-----------------+
| json|
+-----------------+
|[{"a":1},{"a":2}]|
+-----------------+
```
Also, this PR proposes to rename from `StructToJson` to `StructsToJson ` and `JsonToStruct` to `JsonToStructs`.
## How was this patch tested?
Unit tests in `JsonFunctionsSuite` and `JsonExpressionsSuite` for Scala, doctest for Python and test in `test_sparkSQL.R` for R.
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#17192 from HyukjinKwon/SPARK-19849.
## What changes were proposed in this pull request?
Passes R `tempdir()` (this is the R session temp dir, shared with other temp files/dirs) to JVM, set System.Property for derby home dir to move derby.log
## How was this patch tested?
Manually, unit tests
With this, these are relocated to under /tmp
```
# ls /tmp/RtmpG2M0cB/
derby.log
```
And they are removed automatically when the R session is ended.
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16330 from felixcheung/rderby.
## What changes were proposed in this pull request?
Add "experimental" API for SS in R
## How was this patch tested?
manual, unit tests
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16982 from felixcheung/rss.
## What changes were proposed in this pull request?
Since we could not directly define the array type in R, this PR proposes to support array types in R as string types that are used in `structField` as below:
```R
jsonArr <- "[{\"name\":\"Bob\"}, {\"name\":\"Alice\"}]"
df <- as.DataFrame(list(list("people" = jsonArr)))
collect(select(df, alias(from_json(df$people, "array<struct<name:string>>"), "arrcol")))
```
prints
```R
arrcol
1 Bob, Alice
```
## How was this patch tested?
Unit tests in `test_sparkSQL.R`.
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#17178 from HyukjinKwon/SPARK-19828.
## What changes were proposed in this pull request?
Port Tweedie GLM #16344 to SparkR
felixcheung yanboliang
## How was this patch tested?
new test in SparkR
Author: actuaryzhang <actuaryzhang10@gmail.com>
Closes#16729 from actuaryzhang/sparkRTweedie.
## What changes were proposed in this pull request?
RandomForest R Wrapper and GBT R Wrapper return param `maxDepth` to R models.
Below 4 R wrappers are changed:
* `RandomForestClassificationWrapper`
* `RandomForestRegressionWrapper`
* `GBTClassificationWrapper`
* `GBTRegressionWrapper`
## How was this patch tested?
Test manually on my local machine.
Author: Xin Ren <iamshrek@126.com>
Closes#17207 from keypointt/SPARK-19282.
### What changes were proposed in this pull request?
Observed by felixcheung in https://github.com/apache/spark/pull/16739, when users use the shuffle-enabled `repartition` API, they expect the partition they got should be the exact number they provided, even if they call shuffle-disabled `coalesce` later.
Currently, `CollapseRepartition` rule does not consider whether shuffle is enabled or not. Thus, we got the following unexpected result.
```Scala
val df = spark.range(0, 10000, 1, 5)
val df2 = df.repartition(10)
assert(df2.coalesce(13).rdd.getNumPartitions == 5)
assert(df2.coalesce(7).rdd.getNumPartitions == 5)
assert(df2.coalesce(3).rdd.getNumPartitions == 3)
```
This PR is to fix the issue. We preserve shuffle-enabled Repartition.
### How was this patch tested?
Added a test case
Author: Xiao Li <gatorsmile@gmail.com>
Closes#16933 from gatorsmile/CollapseRepartition.
## What changes were proposed in this pull request?
Added checks for name consistency of input data frames in union.
## How was this patch tested?
new test.
Author: actuaryzhang <actuaryzhang10@gmail.com>
Closes#17159 from actuaryzhang/sparkRUnion.
## What changes were proposed in this pull request?
Add column functions: to_json, from_json, and tests covering error cases.
## How was this patch tested?
unit tests, manual
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#17134 from felixcheung/rtojson.
Update R doc:
1. columns, names and colnames returns a vector of strings, not **list** as in current doc.
2. `colnames<-` does allow the subset assignment, so the length of `value` can be less than the number of columns, e.g., `colnames(df)[1] <- "a"`.
felixcheung
Author: actuaryzhang <actuaryzhang10@gmail.com>
Closes#17115 from actuaryzhang/sparkRMinorDoc.
## What changes were proposed in this pull request?
The `[[` method is supposed to take a single index and return a column. This is different from base R which takes a vector index. We should check for this and issue warning or error when vector index is supplied (which is very likely given the behavior in base R).
Currently I'm issuing a warning message and just take the first element of the vector index. We could change this to an error it that's better.
## How was this patch tested?
new tests
Author: actuaryzhang <actuaryzhang10@gmail.com>
Closes#17017 from actuaryzhang/sparkRSubsetter.
## What changes were proposed in this pull request?
This is a follow-up PR of #16800
When doing SPARK-19456, we found that "" should be consider a NULL column name and should not be set. aggregationDepth should be exposed as an expert parameter.
## How was this patch tested?
Existing tests.
Author: wm624@hotmail.com <wm624@hotmail.com>
Closes#16945 from wangmiao1981/svc.
## What changes were proposed in this pull request?
SparkR ```approxQuantile``` supports input multiple columns.
## How was this patch tested?
Unit test.
Author: Yanbo Liang <ybliang8@gmail.com>
Closes#16951 from yanboliang/spark-19619.
## What changes were proposed in this pull request?
Add coalesce on DataFrame for down partitioning without shuffle and coalesce on Column
## How was this patch tested?
manual, unit tests
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16739 from felixcheung/rcoalesce.
## What changes were proposed in this pull request?
Linear SVM classifier is newly added into ML and python API has been added. This JIRA is to add R side API.
Marked as WIP, as I am designing unit tests.
## How was this patch tested?
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: wm624@hotmail.com <wm624@hotmail.com>
Closes#16800 from wangmiao1981/svc.
## What changes were proposed in this pull request?
Fix a bug in collect method for collecting timestamp column, the bug can be reproduced as shown in the following codes and outputs:
```
library(SparkR)
sparkR.session(master = "local")
df <- data.frame(col1 = c(0, 1, 2),
col2 = c(as.POSIXct("2017-01-01 00:00:01"), NA, as.POSIXct("2017-01-01 12:00:01")))
sdf1 <- createDataFrame(df)
print(dtypes(sdf1))
df1 <- collect(sdf1)
print(lapply(df1, class))
sdf2 <- filter(sdf1, "col1 > 0")
print(dtypes(sdf2))
df2 <- collect(sdf2)
print(lapply(df2, class))
```
As we can see from the printed output, the column type of col2 in df2 is converted to numeric unexpectedly, when NA exists at the top of the column.
This is caused by method `do.call(c, list)`, if we convert a list, i.e. `do.call(c, list(NA, as.POSIXct("2017-01-01 12:00:01"))`, the class of the result is numeric instead of POSIXct.
Therefore, we need to cast the data type of the vector explicitly.
## How was this patch tested?
The patch can be tested manually with the same code above.
Author: titicaca <fangzhou.yang@hotmail.com>
Closes#16689 from titicaca/sparkr-dev.
## What changes were proposed in this pull request?
After SPARK-19464, **SparkPullRequestBuilder** fails because it still tries to use hadoop2.3.
**BEFORE**
https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/72595/console
```
========================================================================
Building Spark
========================================================================
[error] Could not find hadoop2.3 in the list. Valid options are ['hadoop2.6', 'hadoop2.7']
Attempting to post to Github...
> Post successful.
```
**AFTER**
https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/72595/console
```
========================================================================
Building Spark
========================================================================
[info] Building Spark (w/Hive 1.2.1) using SBT with these arguments: -Phadoop-2.6 -Pmesos -Pkinesis-asl -Pyarn -Phive-thriftserver -Phive test:package streaming-kafka-0-8-assembly/assembly streaming-flume-assembly/assembly streaming-kinesis-asl-assembly/assembly
Using /usr/java/jdk1.8.0_60 as default JAVA_HOME.
Note, this will be overridden by -java-home if it is set.
```
## How was this patch tested?
Pass the existing test.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#16858 from dongjoon-hyun/hotfix_run-tests.
## What changes were proposed in this pull request?
This pull request adds two new user facing functions:
- `to_date` which accepts an expression and a format and returns a date.
- `to_timestamp` which accepts an expression and a format and returns a timestamp.
For example, Given a date in format: `2016-21-05`. (YYYY-dd-MM)
### Date Function
*Previously*
```
to_date(unix_timestamp(lit("2016-21-05"), "yyyy-dd-MM").cast("timestamp"))
```
*Current*
```
to_date(lit("2016-21-05"), "yyyy-dd-MM")
```
### Timestamp Function
*Previously*
```
unix_timestamp(lit("2016-21-05"), "yyyy-dd-MM").cast("timestamp")
```
*Current*
```
to_timestamp(lit("2016-21-05"), "yyyy-dd-MM")
```
### Tasks
- [X] Add `to_date` to Scala Functions
- [x] Add `to_date` to Python Functions
- [x] Add `to_date` to SQL Functions
- [X] Add `to_timestamp` to Scala Functions
- [x] Add `to_timestamp` to Python Functions
- [x] Add `to_timestamp` to SQL Functions
- [x] Add function to R
## How was this patch tested?
- [x] Add Functions to `DateFunctionsSuite`
- Test new `ParseToTimestamp` Expression (*not necessary*)
- Test new `ParseToDate` Expression (*not necessary*)
- [x] Add test for R
- [x] Add test for Python in test.py
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: anabranch <wac.chambers@gmail.com>
Author: Bill Chambers <bill@databricks.com>
Author: anabranch <bill@databricks.com>
Closes#16138 from anabranch/SPARK-16609.
## What changes were proposed in this pull request?
The names method fails to check for validity of the assignment values. This can be fixed by calling colnames within names.
## How was this patch tested?
new tests.
Author: actuaryzhang <actuaryzhang10@gmail.com>
Closes#16794 from actuaryzhang/sparkRNames.
## What changes were proposed in this pull request
When Kmeans using initMode = "random" and some random seed, it is possible the actual cluster size doesn't equal to the configured `k`.
In this case, summary(model) returns error due to the number of cols of coefficient matrix doesn't equal to k.
Example:
> col1 <- c(1, 2, 3, 4, 0, 1, 2, 3, 4, 0)
> col2 <- c(1, 2, 3, 4, 0, 1, 2, 3, 4, 0)
> col3 <- c(1, 2, 3, 4, 0, 1, 2, 3, 4, 0)
> cols <- as.data.frame(cbind(col1, col2, col3))
> df <- createDataFrame(cols)
>
> model2 <- spark.kmeans(data = df, ~ ., k = 5, maxIter = 10, initMode = "random", seed = 22222, tol = 1E-5)
>
> summary(model2)
Error in `colnames<-`(`*tmp*`, value = c("col1", "col2", "col3")) :
length of 'dimnames' [2] not equal to array extent
In addition: Warning message:
In matrix(coefficients, ncol = k) :
data length [9] is not a sub-multiple or multiple of the number of rows [2]
Fix: Get the actual cluster size in the summary and use it to build the coefficient matrix.
## How was this patch tested?
Add unit tests.
Author: wm624@hotmail.com <wm624@hotmail.com>
Closes#16666 from wangmiao1981/kmeans.
## What changes were proposed in this pull request?
The `coefficients` component in model summary should be 'matrix' but the underlying structure is indeed list. This affects several models except for 'AFTSurvivalRegressionModel' which has the correct implementation. The fix is to first `unlist` the coefficients returned from the `callJMethod` before converting to matrix. An example illustrates the issues:
```
data(iris)
df <- createDataFrame(iris)
model <- spark.glm(df, Sepal_Length ~ Sepal_Width, family = "gaussian")
s <- summary(model)
> str(s$coefficients)
List of 8
$ : num 6.53
$ : num -0.223
$ : num 0.479
$ : num 0.155
$ : num 13.6
$ : num -1.44
$ : num 0
$ : num 0.152
- attr(*, "dim")= int [1:2] 2 4
- attr(*, "dimnames")=List of 2
..$ : chr [1:2] "(Intercept)" "Sepal_Width"
..$ : chr [1:4] "Estimate" "Std. Error" "t value" "Pr(>|t|)"
> s$coefficients[, 2]
$`(Intercept)`
[1] 0.4788963
$Sepal_Width
[1] 0.1550809
```
This shows that the underlying structure of coefficients is still `list`.
felixcheung wangmiao1981
Author: actuaryzhang <actuaryzhang10@gmail.com>
Closes#16730 from actuaryzhang/sparkRCoef.
## What changes were proposed in this pull request?
This affects mostly running job from the driver in client mode when results are expected to be through stdout (which should be somewhat rare, but possible)
Before:
```
> a <- as.DataFrame(cars)
> b <- group_by(a, "dist")
> c <- count(b)
> sparkR.callJMethod(c$countjc, "explain", TRUE)
NULL
```
After:
```
> a <- as.DataFrame(cars)
> b <- group_by(a, "dist")
> c <- count(b)
> sparkR.callJMethod(c$countjc, "explain", TRUE)
count#11L
NULL
```
Now, `column.explain()` doesn't seem very useful (we can get more extensive output with `DataFrame.explain()`) but there are other more complex examples with calls of `println` in Scala/JVM side, that are getting dropped.
## How was this patch tested?
manual
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16670 from felixcheung/rjvmstdout.
## What changes were proposed in this pull request?
With doc to say this would convert DF into RDD
## How was this patch tested?
unit tests, manual tests
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16668 from felixcheung/rgetnumpartitions.
## What changes were proposed in this pull request?
Add R wrapper for bisecting Kmeans.
As JIRA is down, I will update title to link with corresponding JIRA later.
## How was this patch tested?
Add new unit tests.
Author: wm624@hotmail.com <wm624@hotmail.com>
Closes#16566 from wangmiao1981/bk.
## What changes were proposed in this pull request?
Support for
```
df[[myname]] <- 1
df[[2]] <- df$eruptions
```
## How was this patch tested?
manual tests, unit tests
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16663 from felixcheung/rcolset.
## What changes were proposed in this pull request?
```spark.gaussianMixture``` supports output total log-likelihood for the model like R ```mvnormalmixEM```.
## How was this patch tested?
R unit test.
Author: Yanbo Liang <ybliang8@gmail.com>
Closes#16646 from yanboliang/spark-19291.
## What changes were proposed in this pull request?
spark.lda passes the optimizer "em" or "online" as a string to the backend. However, LDAWrapper doesn't set optimizer based on the value from R. Therefore, for optimizer "em", the `isDistributed` field is FALSE, which should be TRUE based on scala code.
In addition, the `summary` method should bring back the results related to `DistributedLDAModel`.
## How was this patch tested?
Manual tests by comparing with scala example.
Modified the current unit test: fix the incorrect unit test and add necessary tests for `summary` method.
Author: wm624@hotmail.com <wm624@hotmail.com>
Closes#16464 from wangmiao1981/new.
## What changes were proposed in this pull request?
To allow specifying number of partitions when the DataFrame is created
## How was this patch tested?
manual, unit tests
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16512 from felixcheung/rnumpart.
## What changes were proposed in this pull request?
spark.kmeans doesn't have interface to set initSteps, seed and tol. As Spark Kmeans algorithm doesn't take the same set of parameters as R kmeans, we should maintain a different interface in spark.kmeans.
Add missing parameters and corresponding document.
Modified existing unit tests to take additional parameters.
Author: wm624@hotmail.com <wm624@hotmail.com>
Closes#16523 from wangmiao1981/kmeans.
## What changes were proposed in this pull request?
```
df$foo <- 1
```
instead of
```
df$foo <- lit(1)
```
## How was this patch tested?
unit tests
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16510 from felixcheung/rlitcol.
## What changes were proposed in this pull request?
R family is a longer list than what Spark supports.
## How was this patch tested?
manual
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16511 from felixcheung/rdocglmfamily.
## What changes were proposed in this pull request?
SparkR ```mllib.R``` is getting bigger as we add more ML wrappers, I'd like to split it into multiple files to make us easy to maintain:
* mllib_classification.R
* mllib_clustering.R
* mllib_recommendation.R
* mllib_regression.R
* mllib_stat.R
* mllib_tree.R
* mllib_utils.R
Note: Only reorg, no actual code change.
## How was this patch tested?
Existing tests.
Author: Yanbo Liang <ybliang8@gmail.com>
Closes#16312 from yanboliang/spark-18862.
## What changes were proposed in this pull request?
It would make it easier to integrate with other component expecting row-based JSON format.
This replaces the non-public toJSON RDD API.
## How was this patch tested?
manual, unit tests
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16368 from felixcheung/rJSON.
## What changes were proposed in this pull request?
API for SparkUI URL from SparkContext
## How was this patch tested?
manual, unit tests
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16367 from felixcheung/rwebui.
## What changes were proposed in this pull request?
SparkR tests, `R/run-tests.sh`, succeeds only once because `test_sparkSQL.R` does not clean up the test table, `people`.
As a result, the rows in `people` table are accumulated at every run and the test cases fail.
The following is the failure result for the second run.
```r
Failed -------------------------------------------------------------------------
1. Failure: create DataFrame from RDD (test_sparkSQL.R#204) -------------------
collect(sql("SELECT age from people WHERE name = 'Bob'"))$age not equal to c(16).
Lengths differ: 2 vs 1
2. Failure: create DataFrame from RDD (test_sparkSQL.R#206) -------------------
collect(sql("SELECT height from people WHERE name ='Bob'"))$height not equal to c(176.5).
Lengths differ: 2 vs 1
```
## How was this patch tested?
Manual. Run `run-tests.sh` twice and check if it passes without failures.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#16310 from dongjoon-hyun/SPARK-18897.
## What changes were proposed in this pull request?
While adding vignettes for kstest, I found some errors in the example:
1. There is a typo of kstest;
2. print.summary.KStest doesn't work with the example;
Fix the example errors;
Add a new unit test for print.summary.KStest;
## How was this patch tested?
Manual test;
Add new unit test;
Author: wm624@hotmail.com <wm624@hotmail.com>
Closes#16259 from wangmiao1981/ks.
## What changes were proposed in this pull request?
Support overriding the download url (include version directory) in an environment variable, `SPARKR_RELEASE_DOWNLOAD_URL`
## How was this patch tested?
unit test, manually testing
- snapshot build url
- download when spark jar not cached
- when spark jar is cached
- RC build url
- download when spark jar not cached
- when spark jar is cached
- multiple cached spark versions
- starting with sparkR shell
To use this,
```
SPARKR_RELEASE_DOWNLOAD_URL=http://this_is_the_url_to_spark_release_tgz R
```
then in R,
```
library(SparkR) # or specify lib.loc
sparkR.session()
```
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16248 from felixcheung/rinstallurl.
## What changes were proposed in this pull request?
Several SparkR API calling into JVM methods that have void return values are getting printed out, especially when running in a REPL or IDE.
example:
```
> setLogLevel("WARN")
NULL
```
We should fix this to make the result more clear.
Also found a small change to return value of dropTempView in 2.1 - adding doc and test for it.
## How was this patch tested?
manually - I didn't find a expect_*() method in testthat for this
Author: Felix Cheung <felixcheung_m@hotmail.com>
Closes#16237 from felixcheung/rinvis.
## What changes were proposed in this pull request?
In this PR, the document of `summary` method is improved in the format:
returns summary information of the fitted model, which is a list. The list includes .......
Since `summary` in R is mainly about the model, which is not the same as `summary` object on scala side, if there is one, the scala API doc is not pointed here.
In current document, some `return` have `.` and some don't have. `.` is added to missed ones.
Since spark.logit `summary` has a big refactoring, this PR doesn't include this one. It will be changed when the `spark.logit` PR is merged.
## How was this patch tested?
Manual build.
Author: wm624@hotmail.com <wm624@hotmail.com>
Closes#16150 from wangmiao1981/audit2.
## What changes were proposed in this pull request?
Reviewing SparkR ML wrappers API for 2.1 release, mainly two issues:
* Remove ```probabilityCol``` from the argument list of ```spark.logit``` and ```spark.randomForest```. Since it was used when making prediction and should be an argument of ```predict```, and we will work on this at [SPARK-18618](https://issues.apache.org/jira/browse/SPARK-18618) in the next release cycle.
* Fix ```spark.als``` params to make it consistent with MLlib.
## How was this patch tested?
Existing tests.
Author: Yanbo Liang <ybliang8@gmail.com>
Closes#16169 from yanboliang/spark-18326.
## What changes were proposed in this pull request?
Fix reservoir sampling bias for small k. An off-by-one error meant that the probability of replacement was slightly too high -- k/(l-1) after l element instead of k/l, which matters for small k.
## How was this patch tested?
Existing test plus new test case.
Author: Sean Owen <sowen@cloudera.com>
Closes#16129 from srowen/SPARK-18678.
## What changes were proposed in this pull request?
Several cleanup and improvements for ```spark.logit```:
* ```summary``` should return coefficients matrix, and should output labels for each class if the model is multinomial logistic regression model.
* ```summary``` should not return ```areaUnderROC, roc, pr, ...```, since most of them are DataFrame which are less important for R users. Meanwhile, these metrics ignore instance weights (setting all to 1.0) which will be changed in later Spark version. In case it will introduce breaking changes, we do not expose them currently.
* SparkR test improvement: comparing the training result with native R glmnet.
* Remove argument ```aggregationDepth``` from ```spark.logit```, since it's an expert Param(related with Spark architecture and job execution) that would be used rarely by R users.
## How was this patch tested?
Unit tests.
The ```summary``` output after this change:
multinomial logistic regression:
```
> df <- suppressWarnings(createDataFrame(iris))
> model <- spark.logit(df, Species ~ ., regParam = 0.5)
> summary(model)
$coefficients
versicolor virginica setosa
(Intercept) 1.514031 -2.609108 1.095077
Sepal_Length 0.02511006 0.2649821 -0.2900921
Sepal_Width -0.5291215 -0.02016446 0.549286
Petal_Length 0.03647411 0.1544119 -0.190886
Petal_Width 0.000236092 0.4195804 -0.4198165
```
binomial logistic regression:
```
> df <- suppressWarnings(createDataFrame(iris))
> training <- df[df$Species %in% c("versicolor", "virginica"), ]
> model <- spark.logit(training, Species ~ ., regParam = 0.5)
> summary(model)
$coefficients
Estimate
(Intercept) -6.053815
Sepal_Length 0.2449379
Sepal_Width 0.1648321
Petal_Length 0.4730718
Petal_Width 1.031947
```
Author: Yanbo Liang <ybliang8@gmail.com>
Closes#16117 from yanboliang/spark-18686.
## What changes were proposed in this pull request?
It's better we can fix this issue by providing an option ```type``` for users to change the ```predict``` output schema, then they could output probabilities, log-space predictions, or original labels. In order to not involve breaking API change for 2.1, so revert this change firstly and will add it back after [SPARK-18618](https://issues.apache.org/jira/browse/SPARK-18618) resolved.
## How was this patch tested?
Existing unit tests.
This reverts commit daa975f4bf.
Author: Yanbo Liang <ybliang8@gmail.com>
Closes#16118 from yanboliang/spark-18291-revert.
## What changes were proposed in this pull request?
Similar to SPARK-18401, as a classification algorithm, logistic regression should support output original label instead of supporting index label.
In this PR, original label output is supported and test cases are modified and added. Document is also modified.
## How was this patch tested?
Unit tests.
Author: wm624@hotmail.com <wm624@hotmail.com>
Closes#15910 from wangmiao1981/audit.
## What changes were proposed in this pull request?
### The Issue
If I specify my schema when doing
```scala
spark.read
.schema(someSchemaWherePartitionColumnsAreStrings)
```
but if the partition inference can infer it as IntegerType or I assume LongType or DoubleType (basically fixed size types), then once UnsafeRows are generated, your data will be corrupted.
### Proposed solution
The partition handling code path is kind of a mess. In my fix I'm probably adding to the mess, but at least trying to standardize the code path.
The real issue is that a user that uses the `spark.read` code path can never clearly specify what the partition columns are. If you try to specify the fields in `schema`, we practically ignore what the user provides, and fall back to our inferred data types. What happens in the end is data corruption.
My solution tries to fix this by always trying to infer partition columns the first time you specify the table. Once we find what the partition columns are, we try to find them in the user specified schema and use the dataType provided there, or fall back to the smallest common data type.
We will ALWAYS append partition columns to the user's schema, even if they didn't ask for it. We will only use the data type they provided if they specified it. While this is confusing, this has been the behavior since Spark 1.6, and I didn't want to change this behavior in the QA period of Spark 2.1. We may revisit this decision later.
A side effect of this PR is that we won't need https://github.com/apache/spark/pull/15942 if this PR goes in.
## How was this patch tested?
Regression tests
Author: Burak Yavuz <brkyvz@gmail.com>
Closes#15951 from brkyvz/partition-corruption.
## What changes were proposed in this pull request?
* Fix SparkR ```spark.glm``` errors when fitting on collinear data, since ```standard error of coefficients, t value and p value``` are not available in this condition.
* Scala/Python GLM summary should throw exception if users get ```standard error of coefficients, t value and p value``` but the underlying WLS was solved by local "l-bfgs".
## How was this patch tested?
Add unit tests.
Author: Yanbo Liang <ybliang8@gmail.com>
Closes#15930 from yanboliang/spark-18501.