Tod means Fox

Supporting decisions through sound data management

Archive for August, 2007

VFP Boot Camp in October

Aug 29

Posted by Tod McKenna in Events and Trainings

I’m just passing this news along. I received it in an email from vfpupdate.com.

Registration ends on September 14, 2007 for the highly-acclaimed VFP Boot Camp. This event delivers three jam-packed days of solid, hands-on learning. From the fundamentals of VFP development through the very latest VFP 9.0 features, this boot camp is designed to quickly get your VFP skills up to speed. Attendees will receive a 500 page manual in both print and electronic format.

For more information and to register, please visit http://visionpace.com/developereducation.html!

Tags: ,

No Comments

David Woods Learns FoxPro

Aug 28

Posted by Tod McKenna in Software Development

David Woods talks about a recent project where he is connecting to an older legacy FoxPro system using ADO.NET. He lists out 6 “stumbling points“, which include some common errors during the connection process as well as some words of wisdom when writing queries. Actually, the best point is number 6. He writes:

6. This applies to everyone. CODE TO INTERFACES. By coding to the IConnection, ICommand, etc. interfaces my switched back and forth between odbc and oledb took seconds not days.

This is solid advice and should be followed by all OOP developers. Coding to interfaces (and not objects) saves time, simplifies designs, and increases flexibility especially among teams of developers. Thanks David for the advice!

Tags: ,

No Comments

Fact Tables

Aug 27

Posted by Tod McKenna in Data Management

Fact tables represent business measurements in a dimensional model (more info: 1, 2, 3). They form the center of the star schema and act as a hub among dimensions for “slice-and-dice” operations to create interesting many-to-many relationships. Measurement data in a fact table is most useful if it is additive (prices and number of units for example); however, non-additive data (such as a percentage) as well as partly-additive data (a metric that is only additive in certain contexts) can also be useful under the right circumstances.

Granularity is the defined level of detail of a measurement. The more granular, the greater the detail. The granularity of the fact table will help determine its type and application. More detail allows you to aggregate data in different ways. If there is not enough detail, then drilling down into the data won’t be possible. Here are the three basic types of fact tables as defined by their grain:

Transaction-level grain

These fact tables are common and represent a business function at the transaction level, often representing the lowest-possible level of data available. An example would be a line item on an invoice or sales receipt; a visit, or “hit”, to a page on a website; or perhaps a single pitch in a baseball game.

Accumulating Snapshot grain

This type of fact is not as common as the transaction-level fact, but very important, especially if you are interested in tracking the lifetime of a process. For example, you could use an accumulating snapshot to store a product’s movement through the supply chain. Each row in the fact table would contain date foreign keys that represent milestones in the process (product ordered, removed from inventory, shipped, received, etc.). This fact would compliment your transaction-level fact table, giving you the ability to view a process’s performance.

Periodic Snapshot grain

The periodic snapshot, like the type implies, is a view of a business process at the end of a specific interval. A summary of measurement data is taken at consistent intervals, such as by hour, day, week, or month. These measurements can then be used for trend analysis and to test business performance over time, without the overhead of aggregating transaction-level details.

Each of these fact table types have different applications in the warehouse. Transaction grain is most intuitive to business users and queries against these tables tend to be simple. If the level of detail in the transaction fact is fine enough, then a periodic snapshot will simply be an aggregate of this activity. If the grain is not fine enough, then the periodic snapshot will give additional insight into the operations of the business by providing a better and complimentary view of business data. Accumulating snapshots work exceedingly well when tracking a process over time. Product orders are a great example, as they move from point-to-point in the supply chain. Any business process that expands beyond a single moment in time is a candidate for an accumulating snapshot.

Next, I will discuss different types of dimensions, followed by examples using facts and dimensions in a data warehouse.

Tags: , , , ,

8 Comments

Guineu means Fox

Aug 25

Posted by Tod McKenna in Software Development

If you haven’t discovered yet, the Foxpert himself, Christof Wollenhaupt, is engaged in an interesting and important extension to VFP9: Guineu. It is billed as an “alternative runtime library for Microsoft Visual FoxPro® 9.0 that runs on any Microsoft .NET compatible platform”.

Hopefully I’ll be able to take a closer look by participating in some beta testing.

Oh, and yes, according to Christof:

Guineu is Catalan, a language spoken in Northern Spain, Andorra and some parts of France. La guineu (it’s female) translates to “fox” in English. When pronounced the “U” is silent and “E” and “U” are two separate vowels.

…guineu does means fox!

Tags: ,

No Comments

Newton’s Method for Approximating the Zeros of a Real-valued Function

Aug 21

Posted by Tod McKenna in Software Development

Need to find the root of a real-valued function and simple algebra just won’t cut it? Then Newton’s method is for you (if you can remember back to your calculus days, that is).

Newton’s method is a root-finding algorithm and uses iteration to solve nonlinear equations. The process involves making a guess at the true solution and then applying a formula to get a better guess. This process repeats until the (approximate) answer is found. The methodology is part of numerical analysis, which is the “branch of mathematics dealing with methods for obtaining approximate numerical solutions of mathematical problems (dictionary.com)”. Specifically, the type of problems fit into the “continuous mathematics” category that are common in medicine, engineering, science, and business industries. (For a good discussion on continuous verses discrete mathematics, check out this blog entry by Mark C. Chu-Carroll or this page from ESC.EDU.) As you might guess, continuous mathematics is very important in business intelligence, specifically statistical analysis of data.

The general form to find the approximation is as follows:

x_{n+1}=x_n - frac{f(x_n)}{f'(x_n)},!

Here’s how it works: Start with an initial guess (get out your graph paper!). Then the function draws a tangent to approximate the function based on your initial guess. Next, the function computes the x-intercept. This x-intercept will be better than the initial guess, and closer to the answer (the root). It then repeats with the x-intercept as the new guess.

The FoxPro code:

LOCAL lcFunction, lnInitialGuess, lnIterations 
 
*-- set the fuction, my initial guess, and number of iterations
lcFunction = "(1/4)*x^3 - 3*x^2 + (3/4)*x - 2"
lnInitialGuess = 2
lnIterations = 10 
 
*-- run newton
newton(lcFunction, lnInitialGuess, lnIterations )
 
*-- foxpro does newton:
FUNCTION newton
LPARAMETERS tcFunction, tnInitialGuess , tnIterations 
 
LOCAL lcDerivative AS String
LOCAL lnNewGuess, lnIter , x , lnY , lnSlope , lnCol5 , lnCol6
 
CREATE CURSOR crNewton (col_n n(2), col_Xn n(14,5), col_fXn n(14,5), col_f1Xn n(14,5), col_ys n(14,5) , col_ng n(14,5))
 
limh = .000001
lcFunction = ALLT(tcFunction)
lnInitialGuess = EVL(tnInitialGuess,0)
lnIterations = EVL(tnIterations,1) 
 
lnNewGuess = lnInitialGuess
 
*-- Loop for each user iterations
FOR lnIter = 1 TO lnIterations
 
  *-- Set guess
  x = lnNewGuess
 
  *-- Run Function
  lnY = &lcFunction
 
  *-- Find Derivative
  lnHoldx = x
  x = x + limh
  lnFuncpluslimh = &lcFunction
  x = lnHoldx
  lnFuncminlimh = &lcFunction
  lnSlope = (lnFuncpluslimh - lnFuncminlimh) / limh 
 
  lnys = lny/lnSlope 
 
  *-- Find New guess
  lnNewGuess = x - lnys
 
  *-- Populate the cursor
  INSERT INTO crNewton (col_n  ,col_Xn ,col_fXn ,col_f1Xn ,col_ys ,col_ng) ;
                VALUES (lnIter ,x      ,lnY     ,lnSlope  ,lnys   ,lnNewGuess)
 
  *-- Reset the guess and continue
  lnNewGuess = lnNewGuess
 
NEXT
 
SELECT crNewton
LOCATE
BROWSE NORMAL

In this run, with the equation I have defined, the approximation of x is 11.80326. If you plug this number in for x in the original equation, it evaluates to .00014. Pretty close! And that’s exactly what this algorithm is great for. Try these additional functions to get the hang of it. You can change the guess and iterations as you like as well:

  • “x + 1″
  • “x^3-12″
  • “x*(x^3)-(3/4)*x”
  • “((x^4+1)*(x^3+2))-11″

 

Tags: , ,

1 Comment

Introducing the New Whitepaper Section

Aug 19

Posted by Tod McKenna in Personal

I wanted to take a moment to introduce a new section on Tod means Fox: Whitepapers.

This page lists articles, whitepapers, memos, and other documents that address a business need, offer advice, breakdown a process, or otherwise present some technical overview. I was tempted to add these items to my Resources page, but in the end determined that separating articles from resources was a good idea. I generally define a resource (such as the FoxPro Wiki) as a conglomerate of information.

Currently, the items on this page were pulled from my Firefox bookmarks and are subject to change.

My hope is that you will have some good suggestions on what should be linked here. Contact me (privately via email at tod at grengama dot com), or by commenting to this post with your suggestions. The only rule is that when a reader clicks on a link, there is no registration or sign-up process: just the article. It would be even better, if you are a copyright holder, to let me host the article or whitepaper on my server — repackaged and refreshed.

I hope you all find this new page useful.

Tags: ,

No Comments

FoxShow 41 FoxForward (and beyond)!

Aug 18

Posted by Tod McKenna in Events and Trainings

I wanted to take a moment to draw your attention to an interview on Andrew MacNeill’s FoxShow regarding FoxForward 2007. It’s a few months old, but will certainly give you an idea on what to expect if you head out.

Here is the link to the audio (mp3).

If you want to know more about the conference, or would like to hear Andrew or Kevin’s voice, check it out!

Tags: , , ,

1 Comment