During SQL Server Geeks Annual Summit 2016, #SSGAS2016 I was really impressed by Wee Hong Tok for his session on ‘SQL Server R services’ where I got to know more about SQL server 2016 and R services and the demo he presented in that particular session regarding Telco Customer Churn.
Also, another session that I am really impressed by Jen Stirrup for her session on ‘Delivering Practical Analytics and Results with Cortana Analytics’. And also impressed by Andreas Wolter Amit Bansal ,Other Speakers & SQL Server Geeks Community for such a friendliness and hospitality.
Since, I also work in Telco sector so first thing I want to try it out is what Wee Hong Tok had shown in his demo and thanks to Jen Stirrup where she introduce me with Cortana Analytics, I don’t know much of this SQL Server 2016 with R services and regarding Cortana Analytics now I want to know more on these.
One can find regarding Cortana Analytics and Telco Customer Churn from below links.
So, if you also want to try out this then all you need is firstly download the SQL Server 2016 Developer Edition or SQL Server 2016 Evaluation version for 180 Days.
For step by step installation you can follow below link
Before installation SQL Server 2016, you need to install Java Development Tool kit. http://www.oracle.com/technetwork/java/javase/downloads/jdk8-downloads-2133151.html
Also, don’t forget to check the R Services (In-Database) tab during installation, also there is separate R Server (Standalone) during installation just prefer R Services ( In-Database) option.
You can find more on R Services (In-database) and R Service (Standalone) from below links.
Once installation compete, in order to test Telco Customer Churn just go to below link to get the backup file for Telco Customer churn in GitHub link below,
Get the teloedw2.bak file and restore in SQL Server 2016 database, now read the Read.md file which contains:
Restore the database provided (telcoedw2.bak)
Run the code in TelcoChurn-Main.sql
TelcoChurn-Main.sql - Use this T-SQL script to try out the telco customer churn example.
TelcoChurn-Operationalize.sql - T-SQL scripts to create the stored procedures used in this example.
while going through above scripts and trying out Customer Churn I got few errors, so I am sharing my finding here, anyone if gets same error then this post might be helpful.
After installing SQL Server 2016 developer edition and trying for this Telco Customer churn SQL server scripts. I got certain error while running scripts TelcoChurn-Main.sql and TelcoChurn-Operationalize.sql. For first, after google I found out we need to change the Memory_Limit_percent in order to run above sq. scripts. So, just add MEMORY_LIMIT_PERCENT=50 into the config file for Rlauncher.config which can be found in below location:
C:\Program Files\Microsoft SQL Server\MSSQL13.MSSQLSERVER\MSSQL\Binn
Change as below;
RHOME=C:\Program Files\Microsoft SQL Server\MSSQL13.MSSQLSERVER\R_SERVICES MPI_HOME=C:\Program Files\Microsoft MPI INSTANCE_NAME=MSSQLSERVER TRACE_LEVEL=1 JOB_CLEANUP_ON_EXIT=1 USER_POOL_SIZE=0 WORKING_DIRECTORY=C:\PROGRA~1\MICROS~3\MSSQL1~1.MSS\MSSQL\EXTENS~1 MEMORY_LIMIT_PERCENT=50
While running this script TelcoChurn-Operationalize.sql we need to firstly install the required R packages, I have also uploaded one SQL File which installed required R packages before running the scripts of telco customer churn into my folk repository for sql-server-samples. The file contains enabling sp_execute_external_script to run R scripts in SQL Server 2016 and installing required R packages in order to run the Script TelcoChurn-Main.sql and TelcoChurn-Operationalize.sql successfully. Since we need to install those R packages before we run the scripts in order to avoid the error. https://github.com/maharjananil/sql-server-samples/blob/master/samples/features/r-services/Telco%20Customer%20Churn/SQL%20Server/Enabling%20R%20scripts%20to%20run%20and%20Installing%20required%20R%20Packages.sql
After that all the scripts run successfully and now you can learn more about R Services and R scripts & algorithm used and then test along with your own Telco data. Which I am planning to try out and will try out for sure.
There is always something new to learn into the world of DATA so called BIG DATA & now DATA Science/Analytics and Machine Learning.