Blogs

Wearable Technology – A boon for Big Data Analytics

The computing industry has been on an innovation high speed for over a decade now. Apple’s ability to redefine the appeal of whole categories of computing has managed to gain an unerring faith of millions of people. Apple has managed to popularized most of the existing technologies - with the Macintosh computer in 1984, the iPod in 2001, the iPhone in 2007 and the iPad in 2010. The recent launch of Apple smartwatch is keeping all the technology geeks on an adrenal rush.

With the likes of smartwatch and google glasses, the tech industry has moved into the new trend of ‘Wearable Technology’. So what exactly is this new trend of ‘Wearable Technology’?

Wearable technology is nothing but small electronic devices that can be worn around the body. It was Google Glass that marked the advent of wearable technology and has aroused excitement among the masses. And now, similar technology like smart watches, fitness monitors, shoes, and headgears that is going to rock the tech- world.
What is interesting to see is that, consumers are prepared to spend a reasonable amount of money on this still-nascent technology. A recent report from Credit Suisse showed that wearable technology will be $30 to $50 billion business!

According to a research conducted by PwC’s Consumer Intelligence Series that surveyed 1000 American consumers, tech influencers and business executives, 20% of the adults in the US already own a wearable device. An adoption rate which is on par with that of the adoption rate of tablets in 2012.
In India too, the market for wearable devices could touch a whopping 200 million within the next three years. Have a look at this 2014 study done by economist.com:

Why does this boon in wearable technology interest us?

These wearable devices collect lots of information about their users, and companies want their hands on it. With the growing trend of big data analysis, companies want to learn from the results these devices are recording. Wearable technology will not only enhance consumers' lives, but will also provide a new source of commercially exploitable data.

Wearable technology is making big data even bigger. Generally, the term big data is used to describe the voluminous amount of data a company and its entities create. However, a huge amount of personal data can be collected from many of these devices.

Wearables and handhelds make it possible to collect location data on an individual too. Multiple apps collect location data and can track movements 24x7. Every time one swipes their cards, a lot of digital financial information is also released.

Along with all this, wearables are now coming with apps which can focus on an individual’s health and other personal information. Putting all these pieces together, one can create a detailed profiling of an individual, making every individual a data generator. The amount and granularity of data then received, gives big data another meaning all together. This data can be used by companies for targeted advertising, customer segmentation and other clustering/targeting strategies.

For example, health and fitness gadgets can capture sensitive details about a person’s health, and send it automatically to the cloud for potential processing by the tech vendor, who may then wish to share it with third parties for 'big data' profiling. One may wonder what companies would do with people’s fitness results. Well, for one, they could learn fitness trends and the more popular ways people like to exercise. Gyms could use that information to save money or create tailored programs based on which day it is. They can determine when to have more staff on hand, and when they don’t need to have as many trainers available.

Healthcare is one industry which can benefit a lot with the wearable technology data. Healthcare data is growing at a rate of 35% per year. Smartwatches gather and transmit data in real time, all the time. Such data will allow researchers to look at even minute data points and analyze hundreds of readings per second from thousands of patients and attain a critical mass of data to detect patterns and make new discoveries.

However, this data brings with itself the complexities of data mining and management. Knowing what data to discard and what to keep in the mammoth database asks for extensive data mining. This is known as the Data Landfill problem, which is just getting worse. Data security is another issue which becomes a big concern. Once these problems are taken care of, this data can give rise to some extensive big data techniques like voice based analysis, machine learning etc.

It would be interesting to see if the big data capabilities of the analytics space can keep up with the ever increasing data inflow.

Building clusters in Tableau

Modern day analytics sees Tableau as an efficient tool for Data visualization. But it has a lot more to offer, Over and above the basic visualization, Tableau, when integrated with R can be used to perform data analytics.

Let us look at the concept of clustering in Tableau with the help of K-means algorithm. Clustering is a very important statistical method through which we can find different characteristics in a population, e.g. finding out different type of customers in a retail store. This process of understanding different traits of a variable in a data can certainly be aided by Tableau’s strong visual representation style.

One of the most fascinating features of Tableau is its ability to connect to R dataset and execute different R commands and functionalities. In this blog, let us see how to establish a connection between R and Tableau, and do a deep dive analysis with the help of the perfect synergy between R and Tableau.

This process can be divided into several small processes-
1. Connection between R and Tableau
2. Getting the data into Tableau
3. Building the clusters in Tableau

Sounds easy!! It is…. All right enough said…now let’s get to work…

1. Connection between R and Tableau

  • a. The first step involves installing an important package in R named “Rserve” (http://cran.r-project.org/web/packages/Rserve/Rserve.pdf.)
    We can follow the following steps shown in the snapshot to installing it in R Studio -
  • b.In the next step, we are going to establish the connection and make sure these two are properly connected-

Step 1 is complete once we get the success message popping up through the computer. Now moving into…2nd process…..

2. Getting the data into Tableau

  • a. We are going to work on Iris data available with R. In order to do this, we are going to export the data as a .csv file with the help of following code-
  • b. As we will be using the same data in Tableau, we are going to import the data -
    Once we have imported the dataset into tableau we are going to set off the Aggregate feature available in Tableau-
    Now that we are fully prepared, we have to understand the basic rules of Tableau and R connection-

    • Tableau can only use functions in R, so Tableau cannot access to data.frame in R.
    • Tableau treats R codes as string, so we have to add “ ” for the codes.
    • When we put variables from Tableau to R, we must specify them as .argN (N=1,2,3,..)

    o If the argN is not an aggregated number, we have to write like .argN[1] in the calculated field
    o When we add Tableau’s data to R, it has to be a constant number or an aggregated number (basically SUM())
    This is a basic graph from the available data combining all the metrics together that we can easily plot, now on to our final and third stage

3. Building the clusters in Tableau-

  • a. We want to build several clusters on this data based on several of its characteristics and we will use Kmeans to build these clusters. We will account for all the variables while building the clusters. We will use the SCRIPT_INT function in Tableau to create these clusters-
  • b. Now let’s make this process more flexible by giving users option to select number of clusters. We will create a parameter with the following specifications-
  • c. Now we need to include this parameter in the calculated field “Clusters” –
  • d. Once we have included the parameter in the calculated field, we will show the parameter control for the created parameter “Number of Clusters” and voila , we have our own clusters which we can select-

Reference

1.Data - Iris Dataset available in R

2.Methodology - http://tallman-world.tumblr.com/post/110391396547/tableau-r-how-to-connect-tableau-and-r-and

All Rights Reserved 2015 | www.platinumetrix.com
Designed & developed by AmazAntz