Cubegeek

Obligatory Seriousness on Apple Maps

I made a very stupid mistake yesterday. I forgot to pack my MacBook's power cable  for my 2 day trip to San Francisco. Always in the back of my head, I know that I can pickup socks or even a pair of slacks if I mispack. I realized once I got off the plane that I would need to make a trip to the Apple Store. So I had to find my way there. This was going to be my first test of the new Apple Maps, which was to get me via public transit (BART) from SFO to Market Street. Either way it was going to be cheaper than taking a limo, BART cost $8.50 the limo costs $30 plus tip. 

The signs at the airport are pretty good. Good enough for me to follow arrows to the Airtrain. The Airtrain (free) went to the BART station. I put ten bucks on the card and 30 minutes later, I was at the Powell Street Station, two blocks away from my final destination. 

The problem with Apple Maps is that it doesn't have any information about the BART. However, as soon as you ask for that local information, Apple Maps shows you the door to the Apple Store and a selection of local transportation routing apps. So I had to wait 30 whole seconds to choose and download Embark iBART which cost me, uhm.. free.  I've actually done this routine before, with Google Maps which didn't have bus schedules for Washington DC. I got that app (I think I paid a couple bucks for it) and was satisfied.

My problem was now looking up the price of the new power cable. $79 bucks at the Apple Store. Are you kidding me? So now I use Apple Maps to find a used dealer around here where I can get a second hand cable and save some money. Mission accomplished with Apple Maps. I found Keane over in the Flood Building.  I saved money by getting a used one, not as much as I would have liked. I left that building feeling really pissed off at myself for leaving my power cable at home. That mistake cost me more than 65 bucks.  Getting a free Apple Maps, saved me $21.50 each way, given the comfort I now have taking BART.

But that's just money. What about technical features? I have a hard time not being a Boy Scout snob about this whole thing. If the difference between Apple Maps and Google Maps leaves you lost and stranded, I would say that the defects in your sense of direction are likely to cost you orders of magnitude more money than the difference you will pay between the two apps. In other words, if Apple Maps isn't good enough for you, then you're more than simply lost, you are simply clueless. 

Does that make me a fanboy? Sure, if it makes you happy to call me that, go ahead. But if you can't get around without turn by turn instructions, what does that make you?

Bucketmaster

Bucketmaster
M. Bowen
November, 1999
---

This is the first in a series of tools to enhance the data-management capabilities of essbase formed by a suite of backend tools which will enable the e-business.

The Bucketmaster will fit into the context of API based tools which work on essbase components. The should be developed outside of the core product as if they were 3rd party tools. They can later be integrated as options enabled by our license.

The purpose of this tool ist to perform a one-off or two-off analysis of a stream of dimensional data. It should allow dbas to quickly bucketize a flat dimension into either a dimensional hierarchy or a virtual attribute dimension. The source for this can be a flat file, or other stream directed to it via the rules editor, or an actual built essbase index, against which the sparsity of a single dimension can be judged. It should also be able to run directly against relational sources.

The overall goal is to give the multidimensional dba the opportunity to optimize design of the essbase database based on real data provided. This analysis should be able to be performed before or after loading data into Essbase.


Part of the Bucketizer is a Tokenizer. This tokenizer should be able to create a surrogate key against a single dimensional source which can be output back into a relational database or flat file. This will allow Essbase to quickly process odd files.

Outputs of the bucketizer will be:

Single Dimension Operations - Against Counts

1. an 'equal bucket' flat line distribution of source keys in the dimension with equivalent sized buckets in a 1 to n distribution.

2. Size constraint distribution of source keys in the dimension. Buckets would all be of equal

3. Bell Curve distribution of source keys in the dimension.


Single Dimension Operations - Against Measures

or the same thing against a second dimension.

2. A bell curve distribution in the form of a dimensional hierarchy or virtual attribute dimension.

3. An 80/20 (adjustable) distribution of dimensional data against a pre-defined or dynamic measure.

Dual Dimension Operations - (Sparsity Managment)

4. a graphical output of the distribution of data within an essbase cube for a two dimensional combination, or the same against a relational source. In this regard I am envisioning a 2dimensional field with red dots signifying the hits within the field. This needn't be done by measure but, color could represent a third dimension. As well it could be done in 3 dimensions which could add some value as a visualization tool.

Plots & Stars

By the end of this year, I'm going to be confident enough with APIs and Ruby gems and all the other weirdness I have begun mastery of to start instantiating some of my infantile ideas. As such I may as well start pollinating them here.

In the meantime what I'm doing related to The Industry is the following. I'm building my own embedded alternatives to DIM, DRM, EPMA, FDQM and ODI. That sounds like a tall order but first thing's first it's all about understanding multidimensional models. It is something I understand. So I'm making baby steps in making an everything to everything translator. It would be nice, for example, to have something that could read Essbase OTL files offline, but the current Outline Extractor doesn't do that. Nevertheless, it does turn online outlines into XML, and XML ladies and gentlemen is the gateway drug. 

Since I'm doing Ruby, the first persistent formats I know are YML and JSON. Funny, I thought JSON was a lot more sophisticated than it is. But it's not, it's nice and simple and all that. So while I thought several months ago that I would be interested in using JSON as some super universal data format, I can see that putting structured relational stuff in UTF-8 pipe delimited is really what I like. For persisting data structures, YML is great. I am at the vaguely dangerous level of XML - so far so so. It seems like a lot of trouble to go through, but with a bit of translational chops, I'm working it out. 

So part of my current project is to build a dimension management / workflow integration tool for HFM. Fortunately, HFM spits and sucks XML for all of its metadata. Plus Star Analytics' Command Server can command HFM to spit or suck remotely. Sweet. All I have to do is write your basic operations add, delete, change, search, validate and make the thing XML literate. Then later on I'll add EPMA literacy and then maybe CSV and all that. No doubt I'll reconcile HFM dimension XML with the Essbase XML of the Olap Underground and I will be well on my way to having a general purpose dimension management system.

The purpose of the general purpose dimension management system is obvious from my career-long ambitions which is to leverage all the kinds of models I've worked with over the years. I can then enable best of breed tool selection with the infrastructure middle tier that is the thorax of all the legs of Full360's soon to be world globbing arachnid. All driven with open source software which will be the kind of serious club in the hands of the appropriate cavemen, and appropriately impenetrable for tree-wrapping IT Porsche drivers waiting to jump ship in accounting departments across the country.

--

Anyway here are the two ideas that are implied in the title. 

Plots. 
There's this. I keep losing it. {morphology, canonical, plot, catalog}  It's a catalog of all of the plots of all the fictions of all time. By the time you get to be my age and you can remember every significant commercial jingle since 1972, you start wanting to quit wasting 90 minute blocks watching drama that you've already figured out.  So yeah I'm going to start reading Drieser and Proust and Mann and Goethe and all them. However, I think there's something necessary for Sci-Fi and comics. It's a vocab. I mean, how could you describe a book in ways that catalog them by complexity and plot? That's the idea. Crowdsource literary criticism by plot type, # of characters etc. That becomes the skeleton of of a very cool database. 

 

Stars.
OPML is something that seems to be have been abandoned by the blogosphere. But there used to be a time before twitter when you could keep track of what people were keeping track of simply by subscribing to their Google Reader. Typepad uses it. So whernever I star something on Google reader, it shows up. I need to get into that so that my OPML or whatever goes where I need it to go. I'm tired of waiting for someone else to do it for me. 

I have about 3 or 4 sets of subscriptions - because I read too much news, but also because I can't force the people I'm interested in following to get onto Twitter or Google+ or Blog or Facebook. So I get them pretty much the only place that aggregates all of those subscriptions - which is Flipboard, but then getting stuff from Flipboard isn't easy either. 

So it's a get it together app and the end destination will be Evernote, because it really gets everything.