Data Diving with Mathematica
Wolfram Research hosts lots of popular websites, including Wolfram|Alpha and the Wolfram Demonstrations Project, and we collect a lot of web traffic data on those sites to make sure you, our visitors, are meeting your goals. To really dive deep into that data, our corporate analysis team has built on a number of Mathematica‘s standard data analysis features to develop a powerful, in-house computable data function for studying web traffic and other business data.
In this video, corporate analysis team lead David Howell describes how using Mathematica gives his team huge advantages in discovering new patterns and relationships within our web traffic data and in delivering insightful interactive reports.
Howell says Mathematica is the perfect environment for data analysis because he can apply powerful analysis functions and create state-of-the-art visualizations all in one system. “If we were working with just a statistics package, we probably wouldn’t have this set of graph theory functions or the nice graph plotting functions that Mathematica has. If we were working with something geared just towards nice time series plots, it might not have the same statistical tools as well. But with Mathematica, they’re all right there in the same tool.”
More details and other applications are on the Mathematica Solution for Data Analysis and Mining page. You can find more on Howell’s work in his blog post “Mathematica for Web Analytics.” Other cutting-edge uses of Mathematica are featured in our Portraits of Success.
Thank you for nice video. I really like it
Very interesting video!
Can you point us to any Mathematica code that can analyze a web page and return (among other things) the size (in bytes) and type of object for each object on the page?
I.e. for this page it might return 1 doc object of 53.7k, 2 js objects (1.5k and 8.8k), 1 css object (8.8k) 3 flash objects (1k, 0.9k and 1k), etc.
Very nice and informative article.
In most spheres I believe (as social science, business, trade) there is always a gap between DB awailable & working (reasonable) model. But i believe stricktly mathematical analysis is hardly possible to combine both for there are always too many incoming non-evident (or too evident) factors which it^s always hardly possible to formalize at least in ordinary linear sense. I believe so to say multifolded schemes mulidimensional worth testing but surely I believe creative thought of expert analyst is prior.