Today we are thrilled to announce Metamarkets’ most significant release to date. This is now available to all of our customers!
This new release includes two new visualizations, an entirely new Discovery Feed, the ability to download a complete dimensional breakdown of your data in CSV and a significant number of minor bug fixes and improvements.
Let’s see what the new features are in more detail.
The Discovery Feed: what should I be looking at?
“Robust Principal Component Analysis (RPCA)” is the name of an algorithm used to reveal the internal structure of a dataset in the way that best explains the variance in the data. A few months ago Nelson Ray introduced RPCA on our blog (see Algorithmic Trendspotting and Boy Bands and Football Fans), exploring how it could be used to look for spikes and anomalies in a data set, as well as to identify the unexpected and interesting areas for further exploration.
The power of RPCA is now available to you through our new analytical engine that powers the Discovery Feed. Located in the “Home” section of your dashboard, the discovery feed highlights interesting trends and events in your data; as you can see in the image, the Discovery Feed on our Wikipedia Stream is capable of highlighting the principal news events of the last days by “noticing” increased activity in the “Edits” metric.
Just as in the case of our demo data feed from Wikipedia, statistically significant changes in your recent data are now highlighted in your Discovery Feed, and one simple click on any story will allow you to jump to the relevant data in the Explore section for further analysis.
New visualizations: Bar Chart and Scatter Plot
In line with our mission to assist you in making data-driven decisions, we have added new graphing capabilities to give you more ways to get insights from your data. Let’s take a look at them:
Widely used in all sorts of business contexts, the bar chart is a way of summarizing a set of categorical data using bars of the same height, each of which represents a particular category. The length of each bar is proportional to a specific aggregation (e.g., to the sum of the values in the category it represents).
This type of chart can be very useful in spotting outliers (outliers are values that are markedly smaller or larger than most other values in the same data set) or differences between categories.
Scatter plots are used to plot data points on a horizontal and a vertical axis in an attempt to show how much one variable is affected by another, and are generally used to find a relationship between two variables. This relationship is called “correlation”: if the markers tend to form a straight line in the scatter plot, the two variables have a high correlation. It is important to be careful when evaluating a scatter plot, as sometimes even though a correlation may seem to be present, both variables might be related to some third variable or pure coincidence (correlation does not imply causation!).
The new visualization also offers a few helpful features to make analyzing the data easier. Every data point is shown as a couple of circles, one a light grey one representing the value at the beginning of the selected time period (day, week, month or custom), a dark grey one representing the current value and a dotted arrow to highlight the variation between the two values (if any).
Moreover, the data in the graph is also displayed in tabular form beside it. Hovering over a row will highlight the data point in the graph with a “bubble” containing the same metrics shown in the table.
Full CSV download
Many of you requested a better CSV export, and today we oblige by providing unrestricted CSV download of a dimensional breakdown of your data (previously, only the top 100 rows were available).
To create the underlying data, from the “Explore” page click on the title bar of a dimensional table to expand it, then click on the “Download” dropdown and select “All”, and your browser will download the file containing the entire breakdown for the displayed period.
All the small things…
Not all the changes we made are quite so big and self evident: we also tweaked the application’s main navigation by renaming the “Analytics” section to a more semantically correct “Views”, we started caching data on the client side more aggressively to improve the responsiveness of the UI etc… The fine tuning work required for an application as complex as Metamarkets is never done; because of that we encourage you to report your feedback and suggestions using our feedback widget (the orange tab you can see at the bottom of the screen when you’re logged in), or by getting in touch with us through our support team.
This is just the beginning of a series of updates planned for the upcoming months that will dramatically improve the app and extend its capabilities. We hope you will enjoy this new release just as much as we enjoyed building it for you, and keep an eye out for further announcements very soon.