April 23rd, 2008 by Lino Ramirez
But surprisingly by a larger margin than we expected. The reason behind that is Hillary did extremely well among one group of her core supporters: Catholics (70% of support). Moreover, Obama did not do as well as he should among his core supporters: voters under 30 years of age (62% of support). This is something we indicated yesterday:
“With voters under 30 being the core of his supporters, I think that Obama has a harder task ahead.”
For more info:
The New York Times has a good Profile of the Pennsylvania Primary Voters and the Pennsylvania Primary Results.
Posted in Obama, Clinton, Pennsylvania Primaries | No Comments »
April 22nd, 2008 by Lino Ramirez
Nobody said it better than Obama: “I’m not predicting a win. . . . I’m predicting that it’s going to be close and that we are going to do a lot better than people expect.“
In Pennsylvania, Obama is leading the way in terms of spending (US$ 8.6M to US$ 3.6), number of blog posts, number of Google searches, number of web site visits, etc. On her side, Clinton has managed to stay on top in the polls.
The race in Pennsylvania is an interesting one. Some say that it is too close to call. Others say that demographics will be a defining factor. However, to win in Pennsylvania, Obama and Clinton have to make sure their supporters go to the polling stations. With voters under 30 being the core of his supporters, I think that Obama has a harder task ahead. However, if he and his team pull it off, it would be the upset victory that he needs to ensure his nomination once and for all.
Posted in Collective Intelligence, Obama, Clinton, Pennsylvania Primaries | No Comments »
March 27th, 2008 by Lino Ramirez
What people wrote in Blogs about Kevin Taft and Ed Stelmach during the campaign for the last provincial election? If you wanted to read every single post, you would be on for several sleepless nights. However, by using some Web 2.0 visual representations, in a short amount of time, you could have a broad picture of what was written. The visual representations we used were tag clouds. Tag clouds are generally used to represent tags associated with the content of websites. Nevertheless, you can use tag clouds for many other applications. One of them is to summarize written essays. See for instance the tag cloud below that was made based on Tim O’Reilly’s essay: What is Web 2.0.
The key feature of tag clouds is that the importance of a tag (or word) is shown with font size. The larger the font size, the larger the number of appearances of the word.
Coming back to what people were saying about Ed Stelmach and Kevin Taft, here are two tag clouds. The first one is the tag cloud for what people were writing about Ed Stelmach.
People that wrote post related to Stelmach made emphasis, among other things, on Alberta and Albertans, Vote and election, good and government, and change and Tomorrow. Note that while there were lots of mentions to JOBS there were not many mentions to economy and oil. From the largest Alberta’s cities only Calgary and Edmonton appeared consistently in Blog posts associated with Ed Stelmach. And interestingly, environment is no where to be seen.
Here is the tag cloud associated with blog posts related to Kevin Taft
People that wrote post related to Taft made emphasis, among other things, on ALBERTA and DEBATE, VOTE and election, issues and government, and change and Yesterday. There were many mentions to economy, energy and oil. The four larger Alberta’s cities had some mentions: Calgary, Edmonton, Red Deer and Lethbridge. Interestingly, while environment and environmental issues were discussed constantly by people writing posts related to Kevin Taft, the phrase it’s time did not gain much traction.
To Learn More:
Tools for making tag clouds:
Related Post:
Posted in Alberta, Stelmach, Taft, General Election 2008, Collective Intelligence | No Comments »
March 26th, 2008 by Lino Ramirez
In Alberta’s last General Election, only a few expected the Progressive Conservative Association of Alberta to win in the way they did. At least, that is what the traditional media told us. However, when harnessing the collective intelligence represented by blog posts, it is clear that the Ed Stelmach and the PC Association had a clear edge. When we analyze the number of blog posts for the leading two candidates (Ed Stelmach and Kevin Taft), we have that during the campaign (from February 4th, 2008 to March 2nd, 2008) there were a total of 1,647 blog posts that mentioned either Ed or Kevin1. From those posts, 70% (1,160 posts) were associated with Ed Stelmach while 30% (487 posts) were associated with Kevin Taft. When we analyze the number of votes received by each candidate, we have that the Liberals and the PC s received a combined number of votes equal to 751,890. From those votes, 67% (501,028) were for the PC while 33% (250,862) were for the Liberals. Based on these results, we can hypothesize that number of blog posts create a trusted source of awareness of a brand (in this case the Liberals and the PCs). We will be testing this hypothesis by making predictions for some of the upcoming US Democratic primaries. Please, stay tuned.
1Recovered using Google Blog Search on March 19, 2008
Ed Stelmach’s picture by Chuck Szmurlo
Kevin Taft’s picture by Alberta Liberal Caucus
Key points to take home
- There was a low interest in the election. 1,647 blog posts from more than 3 million Albertans suggest that the parties did a poor job creating awareness, interest, and action for the election
- Awareness is key in an election. Contrary to what some believed (including me), the Liberals did not create enough buzz using Web 2.0 technologies
Posted in Alberta, Stelmach, Taft, General Election 2008, Collective Intelligence | 1 Comment »
January 28th, 2008 by Lino Ramirez
The other day a friend of mine was asking me about possible business areas in which predictive analytics could be used. I told him that the use of predictive analytics was limited only by the imagination of the people using it and the availability of data to create the models. Because he wanted a more concrete answer, I mentioned these examples (I hope they would give you some ideas on how to use predictive analytics in your organization):
- For Financial Institutions:
- Response: Which customers are likely to respond to a specific offer such as a lower interest rate on credit card.
- Cross-sell: Which customers will apply for a mortgage given they already have a credit card with us?
- Up-sell: Which customers will upgrade to a platinum card given that they have a gold card?
- For Retailers:
- Supply Chain Management: What products are sold together?
- Sales: What key factors are associated with predicting our sales? (location, store size, promotions, etc.)
- For Web-based Companies:
- Web site optimization: How are the users in Calgary different from those in Edmonton with regards to how they navigate the site?
- For Cable Companies:
- Churn: Which customers are likely to leave and sign up with another company?
- For Nonprofit Organizations:
- Donor identification: Who will give to us? Who will renew?
- Strategic planning: Are we investing in the right areas to attain our campaign goal?
Posted in Retail, Predictive Analytics, Fundraising | 1 Comment »
January 2nd, 2008 by Lino Ramirez
Predictive analytics or the analysis of current and historical data to make predictions about future events is being used to demystify the process of buying and selling real estate. Redfin, the first online brokerage for residential real estate, is using a branch of predictive analytics (data mining) for lessons on how to sell homes. In the Fall of 2007, Redfin’s computer scientists analyzed data from more than 500,000 visitors to their listing of over 275,000 properties. The main finding of their study is that the primary determinant of how fast a home will sell, and for how much, is the home itself. However, by following seven recommendation that appear in their report, home sellers will yield a small but significant improvement in the results.
Below is a summary of Redfin’s seven tactics for selling your home. For additional details, please, refer to the Redfin’s report.
- Don’t overprice your property to avoid that it stays a long time in the market. The longer a property is in the market the more aggressive the buyers become in negotiating.
- Set your prices to show in web searchers. You need to take into account that buyers usually filter prices in $25,000 or $50,000 increments. For instance, a house priced at $300,000 is likely to be seen more than a house priced $301,000 because the $301,000 home will be excluded for buyers that set $300,000 as their maximum price.
- Debut your advertisement campaign on Friday to maximize the number of viewings during the first week.
- Stay engaged to increase your chances of selling your property faster.
- Market the property online using, for example, craiglist.
- Do not move until you have sold your house to avoid giving the impression that you are anxious to sell.
- Wait until neighboring foreclosures are off the market to avoid that low prices in the foreclosures affect your own pricing.
Posted in Business Analytics, Predictive Analytics, Real Estate | No Comments »
October 16th, 2007 by Lino Ramirez
Super Crunchers? What are they? Super Crunchers is a term used by Ian Ayres to refer to organizations that analyze massive datasets at lightning speed to gain greater insights into human behavior. In other words, Super Crunchers are organizations that are using predictive analytics to gain insights from data.
If you want to have a fresh view on how predictive analytics can help you, I recommend you to have a look at the webcast: Super Crunchers: Why Thinking-By-Numbers is the New Way to be Smart. In this webcast, Ian Ayres, the author of Super Crunchers, tells the secrets of the “Super Crunchers”
From the description of the webcast:
In this brave new world of equation versus expertise, Ayres shows us the benefits and risks, who loses and who wins, and how super crunching can be used to help, not manipulate us. Gone are the days of solely relying on intuition to make decisions. No businessperson, consumer, or student who wants to stay ahead of the curve should make another keystroke without reading Super Crunchers.
Posted in Predictive Analytics | No Comments »
October 9th, 2007 by Lino Ramirez
There are a set of issues you have to consider before getting started.
-
Do you have the leadership commitment in place? It is important that you have the leaders of your organization committed to act based on the results of analytics. No action means no gain. Therefore, you must guarantee that all the decision makers believe in the process and are willing to act according to the recommendations they get.
-
Can you identify an area of your business that could benefit from the use of analytics? You should try to focus in an area that would provide faster results so you can start seeing the benefits of using analytics in the short term.
-
Can you collect the necessary data to do business analytics? Data is King for business analytics. Only with data that captures the essence of the business problem you are addressing, you can expect to succeed in the use of business analytics.
-
Do you have the skills in your organization (or are you willing to get the required skills) to develop and implement business analytics? Outstanding performers can get the most out of even the simplest business analytics tool in the market. On the other hand, poor performers can not shine even with the best business analytics tool in the market.
As you can see, everything starts and ends with people. People are your most valuable asset in implementing an analytics solution.
Once you have all the pre-requisites described above, you can follow this four-phase process to get started with business analytics:
-
Plan by answering the following questions: what business areas can benefit the most from business analytics in the short and long term? What information is needed to measure success? What data is available?
-
Prototype by first choosing an area that will show immediate benefit and that has data available. Second, you must identify the business and technical requirements. Next, you must start collecting and analyzing data to see if any trend or pattern emerges.
-
Select the tools to use. You should try different vendors if possible. The idea here is to combine different solutions to analyze your business problem from different points of view and create an optimal overall solution that can meet a variety of your needs.
-
Implement the business analytics solution in the area of your choice.
As a final note, I want to stress one more time the importance of People in the success of a business analytics implementation. You need the Right People to analyze your data and produce recommendations and you need the Right People to act based on those recommendations. In short: You need the Right People to succeed as an analytics competitor.
Posted in Business Analytics | No Comments »
October 2nd, 2007 by Lino Ramirez
In two previous posts (How predictive analytics can help you. Case 1: Retail and Case 2: Fundraising), I described some of the benefits of implementing predictive analytics solutions. If you read with interest those articles, you might be wondering: how long does it take to implement an analytics strategy and start seeing the benefits? The short answer is: “It depends”. Before embarking on an analytics adventure, you need to identify a part of your business that you want to improve. It is important that you can collect data on that area (do you have the hardware, software, and communications strategy for doing so?). You must also remember that collecting data takes time. For instance, one manager of customer analytics at UPS noted that they collected data for four years before it became usable. The next step is to make sure the leaders of your organizations are on the same page regarding making fact-based decisions. You must also secure and build analytics skills in your organization to ensure smooth “execution” and keep your competitive advantage. Once you have everything in place, you might be looking to several months for a pilot project on a very specific area or several years for an organization-wide analytics strategy.
To conclude, I will leave you an excellent advice offered by Thomas Davenport on his article: “Competing on Analytics”
“The best advice is to begin working on it [an analytics implementation] now, because it typically requires several years for analytical competitive strategies to come to fruition. […] However, despite the difficulty and expense of establishing these capabilities, many of the firms we have identified as early adopters of analytical strategies are clear leaders in their industries. This suggests the time and trouble necessary to become analytical competitors are definitely worthwhile.”
To learn more:
Competing on Analytics by Thomas Davenport, Don Cohen, and Al Jacobson
Posted in Business Analytics, Predictive Analytics | No Comments »
September 25th, 2007 by Lino Ramirez
Donors rarely explicitly reveal their preferences on what non-profit to support. They usually do so implicitly in their events attendance and membership participation. Analysing their behavioral, demographics, geographic, and previous giving data, we can gain insights on what donors want to support, how much they can or are willing to donate, and how they choose between non-profits.
Consider the following example: Suppose a person donates to a children charity. Would this indicate that the person would likely support an educational non-profit? What if the person donates only to one charity consistently? Would this indicate loyalty to that charity? Predictive analytics can help answer those questions by “figuring out” what is on the donors’ mind.
Some Canadian organizations that are using analytics to gain insight on their potential donors and target their fundraising efforts more effectively are Queen’s University, Richard Ivey School of Business at the University of Western Ontario, University of Toronto, and Doctors Without Borders.
Predictive analytics allows organizations to “figure out” what is on the donors’ mind. Using this knowledge to optimize the performance of the fundraising programs is key to raise more revenue for their mission.
To learn more:
Wikipedia’s entry on predictive analytics
Wikipedia’s entry on fundraising
Posted in Business Analytics, Predictive Analytics, Fundraising | No Comments »
