1. df_smf.csv
   Sample output of 1_BatchJob_SMF30Extract.ipynb, which is batch job run time metrics,
   the most important data input for later batch job analytics.
   In the z/OS operation log, SMF provides a common interface to extract system operation
   measurements. SMF Type 30 includes records of batch job operation. Such data could also
   be collected by other third-party software.

2. MasterBatchJob.csv
   Elapsed time of MBJ in one year, sample data simulated for demo.
   The elapsed time of MBJ is measured as the number of minutes between the start time of
   the first job in MBJ and the end time of the last job in MBJ; it can be calculated from df_smf.csv.

3. TxnVolume.csv
   Transaction volumes of various business types in one year, sample data simulated for demo.   

4. calendar_join.csv
   Calendar data with calendar elements like weekday, day, month, etc.

5. widetable_MBJ.csv
   A joined wide table of MBJ elapsed time, transaction volumes, and calendar data, ready
   for model training for 3_BatchJob_MBJ_Prediction.ipynb.

This includes the source code of 4 Python notebooks. You can open them in the Jupyter web environment
and run them one-by-one following the instruction inside the notebook.

0_readme.ipynb
  This notebook provides an overview of the sample project.

1_BatchJob_SMF30Extract.ipynb
  This notebook extracts batch job operation data from SMF data set with Type 30 records.
  You can modify SQL statements in the notebook to extract the information that you’re interested in.

  Please refer to the IBM Knowledge Center about SMF Type 30 records to find out more
  about mainframe job metrics definition according to your z/OS version.
  SMF data set with one year SMF30 data is too large to fit in a sample project so only the
  output from this notebook - df_smf.csv is included for your information.

2_BatchJob_MBJ_DataExploration.ipynb
  This notebook explores MBJ's elapsed time to get insights on trends in timeline,
  correlation to daytime business volumes and periodicity on week/day/month.

3_BatchJob_MBJ_Prediction.ipynb
  This notebook applies three methods to predict MBJ elapsed time based on historical data,
  calendar information, and business transaction volume data, and then emsembles them as final results.

**Note:** In the notebook cell, when reading local dataset on Watson Machine Learning for z/OS, you
will need a token for authetication, the expiration period is configurated by adminitrator, when you
run cell failed and error message shows that token expired, please click on top right button to get a
new up-to-date token to replace the old project context.

This project has 2 flows which you need to open in Watson Machine Learning for z/OS.

4_BatchJob_MBJ_TSPredict.str
  This flow applies a time series algorithm to predict elapsed time.
  
5_BatchJob_MBJ_AnomalyDetect.str
  This flow detects anomalies in elapsed time and business transaction volumes.

Two zip files with same content, BatchJobAnalytics.zip and BatchJobAnalytics.tar.gz, used for
Windows, Mac and Linux, respectively.
In a web browser with IBM Watson Machine Learning for z/OS, you can "Add Project" from file with
the zip file. This will add a new project with the previous Python notebooks and all datasets at one
time. And then you can explore the sample project as you like.

Before you run and explore the demo process of analytics, you can easily find the demo output here:  
4 read-only views of HTML files, with output results of Python notebooks are included.
4 screenshots from flow output, with predicted results and abnormal records detected.