Data science and analytics

In recent years, artificial intelligence (AI) and data science have become trending topics. For example, 84 per cent of decision-makers in IT and management state that the use of AI is a critical competitive factor.

In practice, however, businesses often lack expertise required to identify and implement the right AI use cases from the multitude of data. The main difficulty here is making reliable and strategically relevant forecasts.

Which data-based decisions would you like to make in the future? This question is crucial because in order to generate long-term, successful added value and competitive advantages from your data, you need a data science strategy. Our experts work with you in an interactive workshop to develop the foundation for your individual data science projects and construct the first use cases.

Once a concrete use case has been identified, we will help you conduct the proof of concept and develop a minimum viable product (MVP), regardless of whether it is based on structured or unstructured data. If you decide to go with our adesso Result-as-a-Service offer in this context, we can also implement your entire use case for you.

We also offer you the option of building data science competencies in your company through AI training courses that are individually tailored to your company. These courses provide you with an overview of use cases, necessary skills and technologies. Your data can be used directly in the training courses so that your employees are ideally prepared for what follows. The choice is yours. You get to decide which programming language is best for you and your employees.

1. In the Interaction Room to use case concept

Before the two-day workshop is held, the first step is to identify the existing IT structures, data structures, data quality and information content of the data. IR:analytics combines BI/DWH expertise with your specialist areas and ensures optimal communication and a lively exchange right from the start. During the workshop, the following points will be clarified:

• Understanding
• Identifying
• Ensuring feasibility and viability
• Prioritisation

Based on the results of the workshop, a basic concept will be developed to determine which architecture, technology, data and methodology are best suited for preparing the data and implementing the identified use cases.

2. From concept to prototype with real data

The second step includes the prototyping, with the objective of testing use cases as quickly as possible. This is where the agile, prototypical implementation of the elaborated results begins. Based on prioritised backlogs, a step-by-step implementation is carried out in sprints.

In order to ensure the structuring of tasks and the reproducibility and objectivity of the implementation, adesso combines an agile approach with the CRISP-DM process in data science projects. The process usually goes through several iterations. More data is available after each run, which allows a more targeted procedure for the next run. At the end of the prototyping, the following results are available:

• a description of a scalable architecture for data science and AI
• a prototypical implementation of the use case
• documentation that is transferable to other use cases
• a viewpoint with specific recommendations for next steps

3. From PoC to finished product

Operationalisation involves the go-live of the use case. The third step begins with the development of a scaling plan suitable for big data that allows us to work with you to prioritise the roll-out in terms of the component functions, customers and markets. The roll-out developed from the pilot will be carried out step by step. The number of users will be continually increased to generate more data to optimise the prototype. The optimisation focuses on technical (such as status or uptime) and functional monitoring (such as forecast quality over time). In this way, the pilot is gradually developed into a robust data science product. Extensive testing during the development phase turns the developed pilot into a marketable data science use case.