I, Deven Edward Pitcher, am a Masters Student at Syracuse University, specifically the School of Information Studies ( iSchool ) in the Information Management degree program. I've recently begun to see the appeal of a large salary and a pleasant working environment.

Personally, I have studied web engineering from a diverse interdisciplinary background. I've acquired comprehensive technical skills in math, engineering, and computer science; however, I've paired these disciplines with a passioned study of the human condition. I fell in love with my Bachelors of Science in Cognitive Science from the State University of New York at Oswego, which required a distinct and creative approach to the computational exploitation of the abstract: new ways of representing old problems, framing questions in a new light so that the answers become obvious, and breaking the big ideas into little manageable pieces. I have mental cravings for new ideas, and an insatiable desire to explore farther and develop further.

I sharpen my teeth daily on the very real and concrete issues of web functionality, accessibility, standards compliance, and feature requirements. I write code, I fix bugs, and essentially I solve problems everyday. I build complex websites with cutting edge enhancements. I optimize page load times and ensure that the code I write enables the user, client, or customer to engage in an intuitive and enjoyable experience.

Contact me at: Deven.Pitcher[at]gmail[.com]

MINIMIZE

The inception of this undertaking involves IST 754 Telecommunications and Network Management Capstone, taught by Dr. Milton Mueller.

This capstone serves as the keystone in the bridge connecting academic coursework with the ability to execute professional work of value and merit.

The semester included instruction in the concepts and practices of analytical data modeling. The focus centered on an individual research project in the field of Information Technology, Networking, or Telecommunications.

Categorically, projects were expected to encapsulate Business Plan Modeling, Enterprise-level Design and Cost Modeling, or Public Policy Modeling.

After formulating a significant and valid problem, the task required extensive data acquisition, constructing algorithms to model salient relationships between data points, and generating thoughtful and useful analysis. With these tasks complete, a final report was to be created which would organize the work put forward, and provide outside parties with access to findings in an easily consumable format.

I have taken what I believe to be a unique approach in constructing this web report. It differs from the syllabus description rather significantly, as detailed here...

Stage 5: Final paper
Usually, it is around 40 pages of text, and includes an Excel spreadsheet Model as a separate file.The title page should be clear and include the following items: name of the project;“Final Project in Telecommunications Systems,”your name and email; date. An abstract of 300-600 words should summarize the work performed, its objectives and scope, and the major conclusions. Put page numbers on each page after the title page and abstract. Be sure to properly format footnote/endnotes citations. List and number all bibliographical references at the end of your paper in alphabetical order by surname of author. When referenced in the text, enclose the citation number in square brackets, for example, [1].

My Goals

My true aim and intentions for this course differ from that of the professor's expectation. I am no longer in the Telecommunications and Network Management program, and am using the capstone instead to substitute for coursework in another program. I am actively trying to manipulate the curriculums available to me to craft an academic experience that will best serve my personal goals. The world of technology is ever transforming and currently possesses unparralled posibility for creativity and inventiveness. The role I wish to play in this world is difficult to define primarialy because much of its environment and context is still undiscovered. I am preparing myself for the unknown. Some new breed of technical, entrepenuerial, analyst, evangelist. Titles and job descriptions evaporate and rain down again no more real than the fog we pass through.

It is with these motivations that I have decided to engage in my endeavors in a unique way. Dr. Mueller and I agree that "results will have to speak for themselves".

So my self assigned tasks are as follows.

• To acquire real and live data from Application Progamming Interfaces (api's).
• Build interactive and impressive data models using modern HTML5 programming techniques that will allow my thoughts to live in the browser and be consumed on any device.
• Develop a feature rich web report that makes people glad they're not reading a PDF and an Excel file.
• Perform all technical work with open-source software.
• Provide the source code online for anyone to replicate my process.
• Investigate the e-commerce industry.
• Refine and Demonstrate my approach to analysis and evaluation.
• Turn the entire experience into something of value to me. Make something I can point to and say, "I do good work, pay me lots of money".
• Define an idea that has haunted me for the last 3 years. ( Yuppy/Yuppie )

MINIMIZE

Online retailers, such as Amazon.com, provide a referral sales program which enables independent owners and operators of websites to earn a percentage of all qualifying purchases made when the shopping session originates from the independent website.

The model and analysis proposed herein specifically targets the revenue potential and cost associated with owning and operating a website for the purpose of generating sales referrals. A site of this type is herby refered to as "a yuppy" or "yuppy".

Any potential or current owner/operator of a website can plug in their own yuppy into this modelss data points and help calculate and predict the profitability of their sales referral endeavor. Specifically, the curator of yuppy needs to decide which categories of products to target. Using the amount of traffic and level of interactivity of the site's user base, the site moderator will have to determine which sector of Amazon's payout chart will be most profitable.

E-commerce sites, especially sales referral or "drop ship" sites, are particularly interesting because they require no physical action on the part of the site operator. Inventory, infrastructure, shipping, and customer service is all handled by the actual retailer.

The main focus of a yuppy is to attract site visitors and convert them to customers. Major retailers have this as one of their main tasks as well; however, the operator of a yuppy is willing to bet that they can perform this task just as well, if not better. Perhaps by focusing on unique niches and effectively engaging and communicating with those communities, the operator of a yuppy can find success.

MINIMIZE

This very well might be the most important section of the report. As is often the case, it is how one accomplishes something that is meaningful; to borrow truth from cliche, it is the journey not the destination.

Data Points

These values represent the components of our model which we have available to manipulate to serve our analysis. The following list includes the defintion of the datapoint, where the data can be obtained, and hypothesis to explore with our model.

• Visitors: Number of monthly users; data from site analytics is used
• Customers: Number of users who click through and purchase something. This variable will be a percentage of data point [Visitors].
• magicX: This is the rate at which [Visitors] become [Customers]. Adjusting this is the primary means of realizing profitability and increasing this percentage should be the major goal of a yuppy operator.
• Product Data: This is a list of the 10 most popular products from a variety of Amazon categories. The name, price, and picture are important for this excercise.
• Payout Rate: A sales referral percentage paid back to Yuppy - Amazon has their rates listed at this address. They vary depending on category. Comparing categorical differences could help Yuppy target a specific user-base to maximize profitability.

Tone and Language

The written word is a tool to convey information from the writer to the reader. I prefer to utilize my natural writing style. One that lacks the pomp and circumstance of robotic drone. You will not find use of any majestic third person royal "we". I will speak as myself to you, the reader as if you were an actual human being with whom I wished to share my thoughts with. It could be argued that this demonstrates my lack of knowledge about the audience; however, I would offer that I am considering my audience. I will adapt to whatever writing style my patron or employer prefers. Yet, since in this matter I am working for myself and not some replaceable necktied firm of "Lastname and Lastname", I need not adapt at all.

Application Programming Interfaces

Product data from Amazon.com is utilized for purposes of this excercise. The data was gathered from the Product Advertising API. To target the payout rates for specific categories, we need products from those categories, and that information was provided at this location. Now we are specifically targeting the most popular products because they are proven to be quite purchaseable: that data is available at this location.

To request and parse this data programatically the open-source library ApaiIO from github user Exeu.

With authentication keys from Amazon and effective use of the ApaiIO library, writing some php code yields our data.

foreach ($cats as $catName => $catID) {

	//a search for each category
	$search = new Search();
	$search->setCategory($catName);
	$search->setSort('salesrank');
	$search->setBrowseNode($catID);
	$search->setResponseGroup(array('Large'));
	
	//now we have the xml
	$response = $apaiIo->runOperation($search);

That raw data is rather cumbersome; the response for ten items from a single category yields 3,000 lines of xml data. A sample request is available in the project repo here. Parsed out to a single item gives us this xml data.

Of course this data is now written out to our catdata directory in a format that can be manipulated by front end browser code.

Modeling Tools

From my perspective, HighCharts is the best open-source and free for non-commercial use software to render resizable charts in a web browser. The HighCharts API is extensive and useful. The demo examples provide simple guidelines for quick execution.

Structuring our data for graphing is rather simple, as this sample demonstrates.

$('#percentage_payouts').highcharts({
    chart: {
      type: 'column',
      backgroundColor: "rgba(255, 255, 255, 0.02)"
    },
    title: {
      text: 'Percentage of Referall Payout'
    },
    subtitle: {
      text: 'Source: Associates Program Advertising Fee Schedule'
    },
    xAxis: {
      categories: [
        "Electronics",
        "Amazon MP3",
        "Amazon Instant Video",
        "Game Downloads",
        "Gift Cards",
        "Grocery",
        "Video Game Console",
        "Headphones",
        "Industrial",
        "MyHabbit",
        "AmazonLocal"
      ]
    },
    yAxis: {
      min: 0,
      title: {
        text: 'Percentage Paid'
      }
    },
    series: [{
      name: 'Rate',
      data: [4.00, 5.00, 5.00, 4.00, 6.00, 4.00, 1.00, 6.00, 8.00, 8.00, 6.00]
    }]
  });

To run our model calculations, gather user input, and generate random values when needed, the JavaScript programming language and the jQuery library were utilized. An example of calculating revenue data follows:

function revenueChart () {

    //calculate 12 months of random users
    //calculate 12 levels of customers
    var startNum = Math.floor( Math.random()  * 100) + 100;
    var magicX = .5;
    var customers = Math.floor(startNum * magicX);
    var categories =  [
        "Electronics",
        "MP3Downloads",
        "Video",
        "VideoGames", 
        "Grocery",  
        "Industrial"
      ];//end categories

      var rates = [ .04, .05, .05, .01 , .04, .08 ];//end rates  
      var revenue = new Array(6);
      var yuppyRevenue = new Array(6);
    
    //calculate for each category
    for(var i = 0; i < 6; i++){
     
      var tenItems = window[categories[i]];
      var revenueSum = 0;
      //for each customer
        for(var j = 0; j < customers; j++){
          var rndProduct = Math.floor( Math.random()  * 10);
          var individual = +tenItems[rndProduct].price;
          revenueSum += individual;
        }//end for all users random product purchase
        revenueSum = revenueSum /10;

      revenue[i] = +revenueSum.toFixed(2);
      //we've calculated amazons revenue, lets calculate ours
      yuppyRevenue[i] = revenue[i] * rates[i];
            
    }//end for 6 categories
    

Code Repository and Project Organization

GitHub for programmers is currently the end-all be-all for managing projects. My repo is available at this location. Utilizing the github paradigm provides for local to the machine and cloud copies of the source code. Additionally, the entire version history is preserved by the git protocol. This allows you to roll back through each step in the project and provides the timeline of progress.

Make It Look Easy

If I made it look easy, it's not. I spent a lot of mental cycle ticks uncovering, discovering, doing, re-doing, writing, and bug fixing throughout this process. That's why I felt it was so necessary to document everything. Hopefully the next poor chump that comes along finds something useful here. Then again, suffering alone is fun.

MINIMIZE

As I model yuppy, I begin by modeling individual data points. The rate which Amazon pays to yuppy is different according to category. Modeling these rates allow us to target yuppy to potential higher-yield payouts.

Please click on the [Draw Chart] button below to model the Amazon payout rate per category.

Analyze:

You can see that Industrial is a category with the highest rate of 8%; selecting it for further study seems intuitive. Let's also grab the lowest rate, Video Game Consoles. Since we are testing the yuppy model I decided to include Electronics, Amazon MP3, Instant Video, and Grocery. These were selected because they evened out the category rate range and had robust product data available. Additionally, the Grocery category is rather new and new domains provide interesting opportunties. The digital goods categories were selected for this reason as well.

Product Data

What good are payout rates without product data? The ten most popular items from each category were queried. For each item we have name, price, and an image.

Analyze:

The first thing that should jump out at any competent analyst is the prevalence of $0.00 priced items. Amazon's API did in fact return these items; it is not a fault in my code. Some items might be given away in promotions in conjuction with other products, while some free items are available especially in the digital goods market place.

A clever yuppy operator will then identify the value in carefully selecting products to sell. Wasted effort is usually rather inefficent.

Analytics Data

Google and a host of startups offer site and mobile app analytics data. I would guarantee that every for-profit site in existence employs user data collection. Intelligence agencies, governments, and university IT administrators are all desperately gathering as much user analytics as possible over the web. Needless to say, it's quite valuable.

In today's Internet economy, visitors to a website equal revenue potential. Advertising payout rates are based on numbers of faces and facetime spent eye crawling. Accusations of click fraud and botnet inflated analytics data are quite common.

Yuppy has strength in the fact that it only recievs revenue from actual purchases. "Pay Per Click" is a rather meaningless statistic and yuppy completely ignores it in its business model.

Click the [Draw Chart] button below to generate some randomized yuppy analytics.

Feel free to re-click for different results.

Analyze:

This traffic data was generated pseudorandomly from very simple heuristics that keeps each month close to the other. This simulates a site with a consistent amount of steady visitors. The magicX (rate at which [Visitors] become [Customers]) is set to 500/1000 or .5 or 50%. I would very much love to operate a yuppy with a magicX of 500; however, what creates a high magicX is very much a mystery and is not the target of this report. My opinions on this matter are irrelevant compared to good A/B testing and extensive market research data.

Calculating Revenue

We've successfully acquired and modeled 5 data points so far: Visitors, Customers, magicX, Product Data, and Payout rates. I believe it is time to push the model one step further. Let's take customer data, product data, and payout rate and calculate the revenue yuppy earns for Amazon and what yuppy earns for itself. Click the [Draw Chart] button below.

Analyze:

The number of customers is random, so you're encouraged to re-click for different results. The number of customers purchasing products from each category is identical, so each category is on equal footing in terms of revenue potential. The only variables to effect the differences witnessed is the cost of items within that category and the percentage rate at which Amazon pays yuppy.

With this in mind, we see categories with lots of $0.00 items; MP3, Game Downloads, and Grocery are all very low. Electronics and Industrial have high priced products in their repetoire. Scrolling up, we remember that Electronics has half the payout rate of Industrial. Electronics might well indeed be an excellent target of a successful yuppy.

Note the difference of scale between Amazon and yuppy. When Amazon makes $5,000, yuppy makes $200. There is more revenue for doing the hard work Amazon does, but it also requires more effort. An ambitious yet lazy yuppy operator would seek to obtain the most amount of revenue for the least amount of work.

Profitability

Ok, let's find out if any of this yuppy effort is worth it. Running a yuppy site is not free. Paying technical staff, bandwidth, name registration, and advertising are all likely costs to yuppy operation.

Below I'll ask you, potential yuppy operator, to enter in the number of monthly users you have or expect to have on your site. Let's limit this to 10,000 maximum for purposes of academic exercise.

Additionally, I need a magicX. This determines how many visitors become customers. A magicX of 1000 means all visitors become customers and a magicX of 0 means no visitors become customers.

Finally, enter how much your site costs to operate per month. Max at $1,000.

No symbols please: [$ , .]

Analyze:

With the default values and most other ranges, we see that Electronics, Industrial, and Instant Videos are profitable categories.

It's really up to you, potential yuppy pilot, to determine to what extent this model can influence your decision making. Obviously this is an analytical tool whose intention is to create the important questions for your business model. I've found questions to be more valuable than answers in my experience. Where are the weaknesses in this model? We'll have to conclude.

MINIMIZE

Why Yuppy

This is a materialistic-centered project, the heart of which involves the goal of selling somebody eles's products to Internet strangers. Yuppies (Young Urban Professionals) are a cultural sterotype in which people purchase consumer items in order to assemble their social identity. They're the audience I would target were I to build yuppy.

I could have modeled Bill Gates' efforts to eradicate malaria or some such other more noble endeavor. Instead, I'm providing a model to calculate if it's worth it to hawk wares for a few extra bucks. Maybe I'm the yuppie-scum?

What We Learned

Picking the right products is likely where you will make it or break it. Suiting the products you advertise to your audience would be key: selecting higher-priced items to those who can afford it.

The data point magicX is the focal point of the high traffic model. It directly modifies visitors to customers, and customers are your ticket to revenue. Let's say you run the premier snowboarding community website; you get a prominent member to recommend your yuppy page for snowboard clothing. That reccommendation turns your visitors to customers, that recommendation is your magicX catalyst.

There are likely a myriad of interesting ways to acquire magicX. If you've thought of one, I advise you to exploit it because it is potentially a valuable trade secret.

Big Ideas?

Is there anything monumental here? Sure, there are some solid web practices, decent programming, and some adequate data modeling and analysis - but is it conceptually interesting? meh; on to the next idea, I say.

Value?

Yuppy-Model is an academic excercise, first and foremost. I've acquired more legitimate experience with modern web functionality and potentially demonstrated some useful skills to prospective employers (if they overlook my blase attitude). It's an added bonus if this project nets me a passing grade. ;-)

It's amazing what a competent individual can do armed with a processor and a network connection. The data is out there, in amounts and locations of exponentially increasing magnitude. The effects of big data revolution are imminent. Think what a skilled analyst could do with a couple trillion personal e-mails.

Iterate

If I were to start all over, what would I do differently? Another instance of the question being more important than the answer. Yet I'd likely try to model my other idea: calculating the profitability of reselling local Upstate NY farmers' market goods at a ridiculous upscale to wealthier markets. I think the podunks and hill folk of this area have some high quality merchandise that they part with for $2.50 a quart when NYC, Seoul, Coopenhagen, or London would pay $25.00 a pint. These goods need a website and some marketing; skills that Syracuse University students have in undervalued abundance.

End.

MINIMIZE