At the resent SAS Insight Series Vince and Bill took us through the ins and outs of modernising your SAS platform.
I went along to see what it was all about and to find some gems that I could pass on to my customers.
Vince a 20 year veteran of SAS took us through the current State
Some of the key notes I picked up from Vince’s talk
- Analytics is the hottest market today
- SAS is the number 1 skill to increase you pay check
- Data scientist is the sexist role
- Big data, you have to mention it
- Fly buys card data referenced as big data
- Check out Realme.govt.nz and data.govt.nz
- Hadoop’s driving down the cost
- Text miner is being used to analyse survey results due to the demand to improve customer service.
- Data ages quickly, loosing its value and Hadoop enables you to process it quickly due to its scalabilty
Vince then talked about the analytics life cycle and why you may want to modernise your SAS platform.
- Reducing Risk
- Increasing Size of data
- Faster Speed
- Drive down cost
- Source of innovation
- Improved accuracy
Bill Gibson with over 30 years of SAS experience then took us through the Modernisation process.
Some of the key notes I picked up from Bill’s talk
Modernisation can mean many things and have many drivers
- Business need
- It strategy – sun setting
- new system / process eg. Hadoop
- Attracting new talent / staff
Bill noted that SAS programming is now called SAS script as you don’t need an IT degree to use it.
The pros and cons of using desktop SAS
- Flexibility freedom
- Consistent performance
- Corporate assets on desktop
- Asset reuse
- Data duplication
- Operation jobs running on desktop
Use proc odstext and proc SGPANEL instead of gplot or gtile for graphics.
Bill then took us through the process of
- Creating a Stored process
- Running it through EG and MS
Using DI Studio to import SAS scripts creates the Metadata for you, and includes a nice DI studio job where you can use the lineage to trace you data.
Bill then showed us the Model development process.
- Using RPM via MS Excel to develop the model
- Building the model in SAS DI studio to productionise it
- Store the data in Hadoop so you have the power to process it.
- And using a distributed version of SAS Visual Analytics over Hadoop to visualise the results.
The road show was then finished off with a short panel discussion. Shane.