[SPARK-6083] [MLLib] [DOC] Make Python API example consistent in NaiveBayes

[MechCoder]

No description provided.

[SparkQA]

Test build #28130 has started for PR 4834 at commit 0c5fe03.

• This patch merges cleanly.

[MechCoder]

cc: @mengxr Would you be able to verify this?

[MechCoder]

Hmm. I get an a accuracy of zero for the given example. Not sure where I'm going wrong though :(

[SparkQA]

Test build #28130 has finished for PR 4834 at commit 0c5fe03.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/28130/
Test PASSed.

[MechCoder]

I changed the randomSplit seed and it works better. It should look good now.

[SparkQA]

Test build #28139 has started for PR 4834 at commit 65bbbe9.

• This patch merges cleanly.

[SparkQA]

Test build #28139 has finished for PR 4834 at commit 65bbbe9.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/28139/
Test PASSed.

[mengxr]

@MechCoder Thanks for the update! We only have 6 lines in sample_naive_bayes_data.txt. That's why some random seed would give bad splits.

[MechCoder]

Great. Do you have any more comments?

[mengxr]

We can define a parse function to make the code more readable. Btw, we use 4 space indentation in Python, following PEP8.

defparseLine(line):
parts=line.split(',')
label=float(parts[0])
features=Vector.dense([float(x) forxinparts[1].split(' ')])
returnLabeledPoint(label, features)
data=sc.textFile('data/mllib/sample_naive_bayes_data.txt').map(parseLine)

[MechCoder]

@mengxr fixed !

[SparkQA]

Test build #28152 has started for PR 4834 at commit 1cdd7b5.

• This patch merges cleanly.

[mengxr]

LGTM.

[SparkQA]

Test build #28152 has finished for PR 4834 at commit 1cdd7b5.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/28152/
Test PASSed.

[mengxr]

Merged into master and branch-1.3. Thanks!