Part II - Public Transport - how to get insights on repairs, maintenance, delays and vendors with th
In a previous article you've found the video of the Digital Boardroom I created for public transport. Consolidated insights in maintenance, repair costs, incidents, delays and many more, while being able to access information in real-time onto the transactional level. The article got so much attention, I decided to explain a bit more in detail what is all embedded.
The Digital Boardroom starts with an agenda covering the topics for insights. Each agenda point defines what is being presented on which one of the three Digital Boardroom screens. In my video you've noticed that I unfortunately do not owe 3 touch screens myself, so I simulate in the webbrowser the three screens. On top of this article is a snapshot of the agenda. As you can see in the video, I did some advanced things by bringing my PowerPoint presentation into the Digital Boardroom creating a simple template that allows my to tap from screen to screen. Also notice the collaboration facilities with annotations or the possibility to invite users to your boardroom in real-time . As a quick sum, the key areas of interest for Public Transport are incidents & maintenance management (both device as infrastructure), repair costs and delays.
Data wrangling, blending and Versions
The modeling, wrangling and data enrichment activities are all handled by BusinessObjects Cloud. This website has a wealth of info on it, so I am not going to repeat here. Important for this case is that I anonymized some real data using the BusinessObjects Cloud modeler, but kept the algorithms, and cadence of the data. In total some 12 sources are accessed. The transaction data is up to 30 Million records.
To consolidate the insights that were created using BusinessObjects Cloud, I used extensively the capability of blending different models together. It is here that the underlying Hana Cloud Platform (HCP) pays of. Blending (combining data from different grain) is heavily accessing computer power, and HCP's in-memory computing is a blessing.
But the real cherry is the possibility of loading different version sof the same data. For this boardroom I loaded the Actual data, a Budget-version of the data and a Forecast version of the data. These versions are stored as a separate attribute and behave like a dimension!! Meaning I can create analytics embedding/comparing actuals , budgets and forecasts. Even more, BusinessObjects Cloud allows the user to create versions while analyzing ........ ahhhhh ...... so I can adjust my budget and forecast and SIMULATE the impact !! The graphs are IBCS Hichert compliant too.
Various customer conversations thought me about the key KPI's for managing public transport delays. You'll all see the in the video. In terms of dynamics and interactivity, I went quite far in this Digital Boardroom:
extensive use of calculated measures: the relative percentage KPI's are all calculated on the fly
use of difference-from metrics: to compare dynamically period over period, I used the difference-from calculations
advanced color-coding / tresh-hold: all KPI's are set against tresh-holds (some dynamically calculated) affecting color-indicators
restricted measures: aggregation for certain KPI's is restricted or customized (I.e. only aggregate when certain conditions apply)
Cross-chart and in-chart filtering: data-point or area selections in one chart can affect other
charts or even the whole Digital Boardroom. An example is where the user can use the polygon-selector on a GEO Map to select a certain area that affects a bar-chart with delay-rankings
Machine discovery: I discuss this in a later thread in more detail, but the machine discovery functionality of the Digital Boardroom utilizes the embedded decision-tree predictive algorithm. It allows to find out the correlations of an attribute to certain metrics ...... what do you mean, Iver ... ? ... Well, for example I used machine discovery to learn that only 2 Vendors where "responsible" for 80% of the "serviced-too-late" spare-parts heavily affecting the downtime of certain trains. We might want to adjust our procurement and choose another vendor.
Ranking: dynamic ranking, error lines, variance reporting and reference lines are used in almost any visualization to make them better understandable and more useful.
GEO Spatial analytics is hot, and it is for a reason. Especially when it comes to massive amounts of data, or data objects changing positions as part of your analytics, GEO spatial analytics bring great advantages making the understandable and meaningful. The challenge here is that every public transport devices doesn't follow a standard route from A to B, yet it has various stops at stations in between. Every begin-station, end-station and stop-station has its own Lon/lat coordinates, and as such its individual GEO hierarchies. The support for multi layers of GEO spatial, is a luxury and works perfect.
Very powerful is the boardroom's capability to work with From-To charts: I can plot the data FROM the begin station TO the end station indicating the top 5 worst routes for example. The boardroom also handles dedicated Point Of Interest (POI) maps using Shape files. I found a Shape file on the web that automatically plots the full Belgian railway infrastructure. I also made some POI maps for example indicating the Weather conditions on a day with lots of Train delays.
Planning and Budgeting
I mentioned above, but the possibility to load various versions of the data is really ground-breaking. While working with it, I learned the power. I used input-screens to allow the user of my Digital Boardroom to adjust certain versions themselves allowing for simulation. I also used the automated predictive forecasting feature. Based on a range of time series, BusinessObjects Cloud can predict a forecast or planning. At the end of the video you'll see me predicting the downtime of certain devices by predicting the failure of key components. Of course the aim of this is to adjust the planning for the preventive maintenance so we can avoid the downtime.
Ah, and did I forget something? Yes, I did. I need some more time getting my hands on the Value Tree. It is not yet in the Public Transport Digital Boardroom, but I will do so later on. An excellent blog on the Value Tree is here. An yes, you have already noticed I need to do some minor cosmetics. Not all graphs have their accompanying title; I am a big fan of leaving away anything that can be left out, but some titles are needed.