Week 2 - Analysis Techniques

Data analysis techniques are used to determine how useful data is by inspecting the data, cleaning it, transforming it and modeling it. The information can be used to lead to conclusions and support hypotheses.

Some of the data analysis techniques that are out there will be described briefly:

·         Data Mining

Jason Frand, a professor at UCLA, explains that data mining, which is sometimes referred to as data/knowledge discovery, is a method of analyzing data from different perspectives and using the information from it to reach to conclusions that can be used to increase revenues, reduce costs, etc. (Frand, UCLA) Having a lot of data from different databases can be overwhelming and it can therefore be hard to determine what all the data means. Through data mining, relationships between the different sets of data can be determined and these correlations can be used to make decisions. Frand goes on to outline the 5 steps of data mining:

1.      Extract, transform and load data

2.      Store and manage the data

3.      Provide data access

4.      Analyze the data

5.      Present the data in a useful format

Step 4, the analysis of the data, is where data mining software would come into play. These software also have the ability to use this data to predict behaviors and future trends. My colleague Stephen Bartal discussed how in the past, only humans had the capability to predict behaviors and determine trends. These analysis techniques allow for computers to do this as well. In terms of intelligent building design, it will be possible for these software to be given a BIM or SAP2000 model and expected to determine faults or discrepancies in the design. This would save engineers a lot of time.

·         Predictive Analytics

Predictive analytics is a business intelligence method that uses your purchasing history to predict your behavior. “Predictive analytics’ central building block is the predictor, a single value measured for each customer.” (predictionimpact.com) This predictor can be seen as a form of behavior and can be a field in a database. An example of a predictor would be frequency: how frequently customers come to the store. This means that the data can be sorted in order of frequency, therefore determining which customers come to the store most frequently. Response to marketing techniques will be higher with the more frequent customers.

Having several fields, therefore several predictors can lead to better, more specific results. These predictors can be combined using formulae to achieve more accurate predictions. In terms of Intelligent Building design, predictive analytics can come into play in the construction phase as well as the maintenance phase of the structure’s life cycle. An algorithm can be created that determines what factors would come into play and how they would interplay with each other to cause the replacement of something. This would probably be most helpful for HVAC.

Bibliography

1.      Alexander, Doug. "Data Mining." Data Mining. N.p., n.d. Web. 11 Jan. 2014. <http://www.laits.utexas.edu/~anorman/BUS.FOR/course.mat/Alex/>.

2.      Frand, Jason. "Data Mining: What Is Data Mining?" Data Mining: What Is Data Mining? N.p., n.d. Web. 11 Jan. 2014. <http://www.anderson.ucla.edu/faculty/jason.frand/teacher/technologies/palace/datamining.htm>.

3.       Siegel, Eric. "Predictive Analytics with Data Mining: How It Works." DM Review's DM Direct (2005): n. pag. Prediction Impact. Web. 11 Jan. 2014. <http://www.predictionimpact.com/predictive.analytics.html>.