This is the first out of a series of posts discussing data science at Novulo. Adding intelligence to our components is essential in the world of software engineering as we know it today. We hereby focus on both optimizing components in order to improve the application itself but also on developing high-value components to help customers utilize their potential.

In this first post, we will describe data science techniques improving the application itself. In following posts, the focus will be shifted towards the use of data science and machine learning in order to personalize marketing activities and keeping a customer database as clean as possible serving as a foundation for the most accurate and reliable segmentation, targeting and lifetime value predictions.

In the world of prescriptive low-code software development, applications can easily be composed. Novulo offers over two thousand pre-engineered components enabling you to customize your own application. According to Forrester this is the next step in software delivery. As a result, the options of each application can potentially be endless. Users can design their own environment, processes can be executed in multiple ways and access management is provided on a highly granular level just to name a few examples. And of course, performance is key as is the case with any software application. Customers can manage settings and track performance manually, providing a great level of control but this also requires a lot of resources which can also be put into other activities of the company. So, what if there would be an intelligent way to automate these processes? This is where data science comes to the rescue.

What is data science?

In short, data science is the study of data where insights are extracted from both structured and unstructured data by using different strategies, calculations and algorithms. Data science is typically seen as an interdisciplinary field where IT, mathematics and statistics meet domain and business knowledge. Now we have a definition of data science, the question remains how to apply it within Novulo.

“
Data Science is an interdisciplinary field where IT, mathematics and statistics meet domain and business knowledge.

Novulo collects log data of the application usage resulting in large amounts of data. These data contain patterns describing user behavior which is perfectly suitable for further analysis. By applying data science, user settings can be personalized, processes can be standardized and performance can be monitored with the objective to customize and optimize each application.

FROM IDENTICAL TO PERSONALIZED SETTINGS

A Novulo application typically consists of many pages which can be customized by each user individually. Relevant tiles and panels can be pinned which can differ from one user to the other. In addition, there are grids where each user can select columns which can be filtered and sorted according to the user’s preferences. In short, settings can be arranged in many, many ways and the user often is not aware of the consequences of the selected settings.

“
By analyzing previous users’ behavior, settings can be personalized, resulting in a more efficient way of working.

Some settings might be computationally expensive whereas others will hardly affect performance. Besides performance, settings can also have a huge impact on efficiency. Certain columns can be crucial in the execution of a particular process and pinning a tile or panel might speed up the way you work significantly. A user might learn which settings are relevant as times goes by and experience with the application is gained, but why not use knowledge and insights from previous users’ behavior?

Data science can help us here to link settings to log data so we can find out which settings are popular and which are not. In addition, we can also determine which settings are used more often than others providing another level of information. In order to finetune and personalize the settings even further, users will have to be assigned to particular roles.

TAKING THE SHORTEST ROUTE FROM A TO B

As mentioned earlier, a Novulo application typically offers plenty of options. Users can have access to many pages and, as a result, processes can be executed in multiple ways. By analyzing log data, patterns, describing crucial business processes, can be extracted. This is also known as process mining and has several use cases.

“
The objective of process mining is to extract relevant patterns from user behavior. The resulting processes can be used to optimize the way of working and automate access management.

The first use case is recommending the next step in a process by relating application behavior, the extracted patterns, to business processes. An optimal procedure, or route, can be specified for each process. See it as a navigation system where the application tells you which step is next in order to complete a process.

Next, process mining can be used to distinguish relevant pages from the ones that can be ignored. After all, users will follow different routes to get from A to B, especially in the beginning when they’re not so familiar with the application. Once users are guided to follow more efficient routes and therefore work more efficiently, access to an abundance of pages is not required anymore. So, process mining can also be used to automate access management in an intelligent way.

Finally, the extracted patterns (or processes) can be grouped together in order to be interpreted as roles. These roles can then be assigned to users. An example of a role can be a finance role consisting of several processes. Think of processes as creating an administration summary or making an invoice. Once roles have been assigned to your users, this can be used to improve the settings recommendations by making them more personal. Here, process mining can serve as input for a recommender system. By combining different data science techniques, a third and even better one can be generated.

DETECTING CRITICAL ISSUES NEAR REAL TIME

Being a complete solution, your application most likely hosts critical business processes. Therefore, performance is key in order to work smoothly and efficiently. Investigating network or server-side issues on an individual basis is far from ideal. By applying data science techniques on the log data of your application, every action (including network-, server- and client times) can be tracked. Once aggregated and analyzed, bottlenecks can be detected near real time on a granular level. By using an interactive dashboard, the insights of your choice can be gathered easily.

The data structure at Novulo is organized in such a way that it can be compared across applications. Don’t worry, your user data is safe. Pages, tiles and panels however, can all be linked resulting in even more insights. Wouldn’t it be nice to know that a page in your application is performing differently than it does in others?

DATA SCIENCE, THE OPTIONS ARE ENDLESS…

In this post, some examples have been discussed of how data science can be applied within the Novulo landscape. In other words, how can we add intelligence to our platform in order to improve your application. However, data science can also be applied in many more ways too. In case you’re interested how, please contact us and of course, keep following this blog.

See you next time!

Platform

Solutions

Development

Solutions

Development

{{i.publish_date | date:'d'}}

{{i.publish_date | date:'MMM'}}

{{i.publish_date | date:'yyyy'}}

{{i.description}}

{{i.short_description}}

Solutions