Final Post

 

Final blog post

Hi Everyone, this is my final blogpost. I’m writing this on Thursday October 22^nd and I’ve just finished a rough draft of my final homework assignment. This class has been tough. Like really tough. I’m not really sure why though, I think just the nature of this class is very open ended and perhaps that’s what makes it difficult for me. After all business intelligence is only intelligence for a particular business. No business will have the same intelligence. But perhaps I’m just conjecturing.

First let’s talk about some positives. The structure of this class was really amazing. The weeks flowed together, and I never felt like we we’re jumping around randomly. The content was relevant and not dated. So many classes in undergrad were using information from a decade ago, it was nice to see relevant research.

Next the negatives. Blog posts. Yuck. I can’t find a happy medium between getting my thoughts across and being professional. Am I writing towards my professor or am I writing towards my fellow students? It’s a mess for me personally. I have no idea if my blog posts have been good or bad because there has been no grades or feedback on them. I hope you have enjoyed reading them.

Let’s move onto the class as a whole. I’ll break it down week by week, bringing up points from my last blog posts and expanding upon others. I want to inject my own experience and opinions while also summarizing the content of the module. That is my goal so let’s get started.

 

Module 0:

I starkly remember this week as it was the first week of school and the professor put on multiple 60+ page papers for us to read. Luckily, a lot of that reading was just footnoting and other things and in reality, the reading only took a few hours.

Diving into big data: companies are rapidly plunging into big data. They need to. It’s the only way to stay competitive in the modern market. I worry about the future of small companies who don’t have the resources to have professionals looking at their google analytics. Is the future the end of mom and pop shops? During my original blog post I made an fruit analogy on the Paradigm Shift of business intelligence. How companies are using the whole orange to make orange juice and then sifting out the rind. Instead of the other way around where they discard the rind and then juice the orange. It’s not a perfect analogy but it tracks at-least a little bit. Companies are consuming any and all data they can get ahold of. Looking for patterns. There is no meta-data not worth consuming/recording in the modern day.

This chart shows the amount of data created in each year, and it’s clear to see it’s just going to get larger and larger. This data from Statista.com

Another interesting point I brought up in module 0 was how the lecture from the professor focused on the positives of big data whereas it largely ignored the negatives. To further this point, I’ve included another article written by Slate magazine showing a list of questionably morale and straight up evil organizations using big data. One such example is Baidu, the Chinese search engine that’s actively suppressing search terms that dwell with the pro-democracy protests of 2019. Not all these are big data examples, but some of them certainly are.

Module 1:

This module certainly was a long one. I don’t understand why some of these modules were so short and some were so long. I would of much preferred writing more blog posts on each individual component as I remember this blog was especially hard to write. However, I am proud of this blog post because I remember feeling like I really found my rhythm in this class. I would watch the video lecture, do the additional reading and then go out on the wild web and try to find my own related research to the topic. This allowed me to quickly and easily write a bit of summary, a bit of analysis, and a bit of personal anecdote. The perfect recipe for a blog post… or so I think.

The balanced score card and star schema design are integral to business intelligence and business in general. The star schema design strictly reminds me a database class included in the MIS program. You learn all about the best ways to design a database and a lot of the concepts for the Star Schema Design lecture can be cross applied. I cannot stress how important it is to conceptualize and understand things such as the slowly changing dimensions Before writing software and/or a database. As retrospectively developing for those types of things can be a huge waste of resources. At my current job we’re running into an issue similar to that where we need users to re-register their addresses into our new system. The only problem is that some users are understandably entering a different address than they did previously. Now we have mismatched addresses and have to figure out whether or not to use their old or new address in other tables. It’s a huge mess we’re slowly working through.

There was so much more in this module, such as data quality and Dashboard design and Analysis. I would like to re-point out how wrong Stephen Few was in 2006 about web-based dashboards dying in the future. I to this day still find that prediction very humorous as he could not have been more wrong. There are definite reasons why dashboards continue to be developed using web technologies. Mahipal Nehra does a great job explaining these in his article “Why Businesses Are Migrating to Web Applications”. In his article, his first point is that they are cheaper to build, maintain and they scale well. The dual nature of the web, front-end and backend is a perfect fit for data dashboards.

This was the first blog post after the Dashboard project. A project I really enjoyed. I had never used Tableau before this class and if I take away anything from this class it’s certainly going to be my enjoyment for that software. I just wish that it was free.

 

Module 2:

In this module we learned all about the different parts of web analytics, their meanings and how to interpret them. Specifically, the different types of web traffic including organic, the most desired type of web traffic and paid/sponsored, the least desired type of web traffic. I also learned some new terms such as the idea of Bounce Rate. Ideally you don’t want your users to bounce of your page you want them to stay forever and just keep clicking your ads.  One of the comments left on my initial blog post suggested that I go check the bounce rate for the Google Merch shop, specifically the page where they sell YouTube merchandise. I was able to confirm that users due tend to immediately bounce whenever they see YouTube merchandise, I guess it’s just not the style at the moment. Oh Well, maybe one day.

For my project in this module I used the default Google Merchandise Store and I rather enjoyed seeing how much money google made from its merchandise. I remember coming to quite a few interesting conclusions that I certainly did not have coming into the project. Furthermore, actually diving into the data gave me great insight into the true value of google analytics. There were parts of the google analytics I could have only hoped to cover in my report. I could have fit 5 pages talking about a single product and how it sold in India to people who use Safari in the month of April. My only point being that Google Analytics records a ton of data. Like I mentioned earlier in this blog post. A ton of data is the new standard in business intelligence.

 

Module 3:

This brings us to our final module. The graph and networking module. This module was interesting because it both laid out the simplicities and the complexities of graph theories. It started off by defining what a Node is and what an Edge is. It explained how an edge can be directed or undirected and how an edge can be weighted. All the basic stuff when understanding a network. However, the lectures also dipped its toes into some more complicated graph theory such as graphs that span multiple dimensions. Additionally, this module taught us how to actually analyze and obtain information from a graph. Before this class I had no idea what centrality in a network looked like. Now I know of at least three different types of centralities and what their meaning and use is.

Closeness centrality is the average length of the shortest path between the node and all other nodes in the graph. Betweenness centrality acts a number representing the number of time that node was in the shortest path between two other nodes. Finally, Eigenvector centrality assigns relative scores based on nodes connecting to other high scoring or low scoring nodes within the graph.

Closeness centrality can be used to determine which node is centered in a network, making it a great starting point for a search or similar algorithm.

Betweenness centrality can be used to determine who on a network is critical to communication and if used nefariously could be used to determine which node should be eliminated to best hamper communication. The image below represents the betweenness centrality of a network.

 

Eigenvector centrality can be used to measure the influence of a node within a network. Such as a search engine determining a page rank.

 

Well that’s it from me. Good luck everyone!

 

Sources

Le Bec, Gwendal. “Which Tech Company Is Really the Most Evil?” Slate Magazine, Slate, 15 Jan. 2020, slate.com/technology/2020/01/evil-list-tech-companies-dangerous-amazon-facebook-google-palantir.html.

Nehra, Mahipal. “Why Businesses Are Migrating to Web Applications?” DEV Community, DEV Community, 6 Aug. 2019, dev.to/decipherzonesoft/why-businesses-are-migrating-to-web-applications-1m0o.