Skip navigation
All Places > PI Developers Club > Blog > Authors Ahmad Fattahi

We are drowning in data and starving for information. There are potentially invaluable knowledge and undiscovered relationships hidden in the data already stored in the PI System. The complexity and volume of data makes it formidable for naked eye or classic tools to discover and extract such multivariate relationships.

 


Machine Learning is an umbrella term referring to methods to learn from previous observations and apply the extracted structure for future predictions. For example, I may be able to observe and archive temperature, day of the week, time of the day, season, fuel price, and the going price of electricity in a certain region. Machine learning allows me to deduce a relationship among these variables, build a model, and use it to predict the going price of the market in presence of observations of other variables.

 


As another example I could make observations of various physical variables of my device such as pressure, temperature, viscosity, flow, etc. as well as my device’s status, i.e., failure or functional. All the data is stored in PI System. I can use these past observations to teach my model. Later on I apply my measurements of current physical variables to predict the chances of my device going down. I can then use this result to schedule preemptive maintenance.

 

Do you have other examples that building such a predictive model can help you? We are about to publish a white paper on performing machine learning on PI Data and would love to hear your opinion.

Here is the article talking about the current structure of the giant companies along with this interesting cartoon:

 

2605.Companies-diagrams.png

Ahmad Fattahi

Project Euler

Posted by Ahmad Fattahi Employee Jul 12, 2011

I just heard about this super cool website and problems collection called Project Euler (Thanks to Luis Moux-Domínguez). This is a collection of interesting geeky computer science-type/math problems that you can solve with some analytical as well as programming/scripting effort. I personally found them very mind-stimulating as well as educational.

 

Another cool thing about it is that until you post the right solution it doesn't show you the answer; really daring

 

Enjoy!

 

0334.Problems-Screenshot.jpg

This is vCampus' 1000th day! We are standing here celebrating our great and lively community because of all your contributions and great energy. On behalf of everyone at OSIsoft we would like to thank you for making all this happen. We have proudly represented more than 1800 individuals, and more than 600 organizations via 10,000 posts on our forums. Your success in creating more value out of investment in PI infrastructure is what we strive everyday. Please keep up the great job and make your community thrive!

 

Interestingly enough Thursday June 30 is named "Social Media Day" to celebrate the great effects social media have had on our professional and personal lives in the past few years. As the social place for all PI System developers, architects, system administrators, geeks, and enthusiasts we would like to congratulate you for the great community we have formed together. With best wishes for our community to grow better and more fruitful everyday.

 

As a tangible step to observe this day, we added a "Follow us on Twitter" button to the vCampus pages. By simply clicking on it you can easily follow OSIsoft vCampus on Twitter and stay abreast of the community news; follow us now and be the first to know

 

3060.Twitter-Button.jpg

Ahmad Fattahi

Social Media Day

Posted by Ahmad Fattahi Employee Jun 27, 2011

This Thursday June 30, 2011 is the Social Media Day. As THE online community for the PI System developers we are very pleased and proud to be part of this big picture. We have been great beneficiaries of the great social platform we have formed together on OSIsoft vCampus.

The Internet Corporation for Assigned Names and Numbers (ICANN) just announced an agreement that would expand the generic top-level domains as early as summer 2012.

 

So there will be some new company for the old and familiar .com, .net, .info. and others. Applicants can submit their applications for domains like .coke, .nyc, .tennis, and similar as early as later this year. This could have lots of marketing value for big companies and brands. There can be security benefits as well.

 

All this comes with some hefty cost though. ICANN charges $185,000 per domain application, which typically must include about 150 pages of policy documents. Technical setup takes another $100,000 or so and upkeep can cost an additional $100,000 each year.

 

How about .PI?

Last week Microsoft and Fujitsu announced a date for the very first implementation of the Windows Azure Appliance by Fujitsu. It is coming in August 2011.

 

Windows Azure Appliance is a container pre-configured with hundreds or thousands of machines running Windows Azure platform. Dell and HP are expected to follow Fujitsu soon. The longer vision is that smaller organizations will be given the platform to run their own Cloud.

 

It means the private Clouds are one step closer to reality. This should mitigate security concerns which is one of the biggest issues of hosting large data systems such as PI System.

Ahmad Fattahi

Windows 8 Sneak Peek

Posted by Ahmad Fattahi Employee Jun 2, 2011

Microsoft just released a video and some details previewing the next Windows known as Windows 8. The major UI features that catch one's eye are:

  • It is designed for touch
  • "Apps" are now central pieces in the user experience
  • "Tiles" are replacing icons

To see the video for yourself follow this link.

LinkedIn, the giant professional social networking company, just filed for IPO (became a public company) and raised $9 billion. Zynga, the giant social gaming company, is reportedly preparing to become public soon and raise even more than LinkedIn. Some believe that Facebook, Pandora, Groupon, Yelp, and Kayak will go through the same procedure soon.

 

Do you feel a deja vu here? Is "Social Media" the new .com and the bubble will burst? Or there is enough merit to these new comers that will keep them up where they are right now?

Let's do some data mining! One of the powerful set of features coming with MS SQL Server is data mining. To be more accurate, it comes as part of the SQL Server Analysis Services (SSAS). It comes with SQL Server Enterprise or Data Center. Using that you can build models on top of your data (e.g. a decision tree) and visualize the results for valuable insight into your data. In short, you can turn data into information.

 

For an easy start, we can start with the Excel client built to work with SSAS in a much more handy way. When installed, it adds a "Data Mining" tab to MS Excel making it available to mine data on the Excel sheets as well as external sources.

 

8422.Data-Mining-Tab.jpg

 

There are multiple features and functionalities in the package.

 

 4377.Data-Mining-Tab-_2D00_-Features.jpg

 

To start off with an easy and cool example, let's try to forecast where a PI tag will go next. For that matter, I used PI DataLink to grab the past 5 hours worth of sampled values of the tag Sinusoid on the Excel sheet and called the columns "Timestamp" and "Value".

 

 4834.Sinusoid-Values.jpg

 

Then I clicked on the button "Forecast" from the Data Mining tab in MS Excel. It starts a wizard allowing you to quickly build the forecasting model. All we need to do is to specify the range of the cells containing the timestamps and values (with possibly the column headers) and click Next.

 

 2350.Cells-Range.jpg

 

On the next window we are building the model by defining the inputs (both columns in this case) and indicating the first column as the time stamp. Click Next.

 

 4885.Indicating-inputs.jpg

 

Just accept the defaults on the next window and click Finish. The tool builds the model (using SSAS in the back end) and plots the existing data as well as some (5 by default) steps of prediction.

 

 7610.Prediction-Result.jpg

 

This is how easy it is! By choosing more steps of prediction you can extend the prediction further into the future. Obviously, the farther you get, the less trustworthy the prediction becomes. Also, by choosing the Model tab on the prediction window you can see the details of the predictive model. It can handle more than one independent variable as well.

 

This way, using our PI System, we can easily turn the insight into action by looking into the future!

 

 

Sometimes, organizations face a project that requires some temporary work by a seasoned and knowledgeable expert on PI System. This kind of expertise gets more and more scarce as the project in hand becomes more specialized and involved. Such "freelance" PI developers form a very important part of the vCampus community.

 

If you are an organization with occasional need for such work force, how do you currently go about it? Where do you search? What would be your ideal process to find your expert programmer or system architect? Would you like vCampus to take a more active/explicit role in that direction? For example would you like to see a marketplace for such workforce and expertise?

 

If you are one of those gems yourself who offer their help to other organizations for temporary projects we would like to know you more! How do you see vCampus helping you reach out to your potential customers? Is there anything you would like to see on vCampus to give you more visibility and better reach? How can we leverage the community on vCampus to help this ecosystem evolve?

 

Our goal here, as always, is to help everyone get more value out of their investment in their PI Systems more easily and efficiently. Read the above paragraph as an attempt toward this final objective!

Check out this link from NY Times. Not only it's fun but also it shows a very good example of how you can store a time series event, and then use them for your own predictive analytics in the future. Make sure you click on the "See what the computer is thinking" to see, well, what the computer is thinking! Maybe we can teach kids to use a PI Server this way

 

7635.ScreenShot1.jpg

The answer is it depends! First the easy part: When comparing CPU clockspeeds within the exact same processor family, clockspeed is a good guide to performance because a higher clockspeed means more instructions are completed per second.

 

But when comparing the CPU clockspeeds of different processor designs, it's generally apples-to-oranges. For two CPUs with the same clockspeed, a shorter pipeline has an advantage because it stays full more often. For two CPUs with the same pipeline depth but different clockspeeds, the higher clockspeed gives it an advantage. For two CPUs with different clockspeeds, it depends on the pipeline depth and other factors.

 

For more information and a quick overview of how CPUs work see this article.

Microsoft and Apple are currently two of the few giants in the high tech and computer world sectore obviously. But how do they compare and what is the trend? Can the comparison shed some light on the future direction of our companies? How do our companies compare with those two giants in terms of per employee revenue and profits?

 

Answers to the above questions and much more interesting stats in this article.

 

0257.profit-chart.png

Some screenshots of a pre-beta version of the Windows 8 operating system have just been posted to different websites. One particular feature is the addition of ribbons (the type seen on MS Office products) in different explorer windows. It must have been a very good experience in the Office environment. It may gradually replace the idea of the menu items as we know them today.

 

0131.win8_5F00_ribbon_5F00_01.jpg

 

1830.win8_5F00_m23_5F00_ribbon_5F00_menu2.jpg

Filter Blog

By date: By tag: