Finalise presentation of big data insights

The business needs of an organisation will drive the detail of how the big data insights will be presented. Presentations may be to a conference-style live audience, smaller seminar, or business function groups and could be virtual or pre-recorded.

Format for business needs

The presentation of big data insights needs to meet several requirements. These include:

• Choosing the best way to display data insights 
• Aligning the presentation to business requirements (following policies, procedures and protocols)
• Adapting to the audience (knowledge level and required detail) 
• Appeal to stakeholders to encourage action.

Protocols for communicating and presenting big data insights may include: 

• Organisation brand guidelines and colour schemes to be used in report visualisations
• Includes guidelines on how to share and articulate the revelations or outcomes 
• Examples of storytelling techniques 
• Recommended formats and design principles to present data analysis results 
• Recommendations on presentation style for different audiences 
• Guidelines on how to present non-standard types of analysis outcomes.

Big data visualisation tools

Visualisation tools provide significant benefits when presenting big data insights. These tools are used in the exploration phase to help identify patterns and the presentation process.

Data visualisation tools help present data to make it easy to understand. It can highlight useful information and reduce noise. These tools use visually appealing presentation techniques to show data with appropriate charts and formatting options. This increases engagement and easier interpretation and understanding. Interactive data visualisation increases the understanding of complex relationships hidden in the data. (Deloitte n.d.)

Information graphics, or infographics, are often defined as a visualisation with some data storytelling.

Choice of visualisation tool

An organisation may have procedures to follow when using statistical graphs, plots and information graphics. These procedures may help with tool selection and may include:

• Steps to follow when selecting visualisation types and to configure data input to the visuals 
• Recommended chart types to use when presenting insights from transactional and non-transactional big data 
• Guidelines on how to implement data into visualisation tools.

The following steps can help evaluate data visualisation tools. (Microsoft n.d.)

1. Determine your needs and goals
   1. Ensure you have a clear goal of what you are trying to convey
   2. Document what data you want to communicate 
2. Decide on visualisation features 
   1. Select a tool with flexibility 
   2. Choose an easy-to-use tool 
   3. Ensure there are adequate security features 
   4. Compare functionality 
3. Consider where data will come from 
   1. Determine if the tools can access the organisation’s database format
   2. Choose tools that can match data from different sources 
4. Think about visualisation complexity 
   1. Decide if standard templates are suitable 
   2. Determine required resources for any customisation required 
5. Collaborate 
   1. Choose tools that allow synergy from collaboration 
6. Publication
   1. Ensure tools can publish visualisations as required 
7. Capability 
   1. Determine time requirements based on existing capability within your organisation 
   2. Is local (external) support available for the tool

The choice of visualisation tool was covered in the previous topic. To finalise your presentation, check whether the most suitable visualisation tools have been selected by referring to the previous topic or an alternate checklist of best practices such as this list.

The visualisation tools for presenting big data insight depend on the data category. Big data can be categorised as either structured, semi-structured or unstructured. A definition for each category and the types of presentation suited to each category is covered in the previous topic.

Visualising unstructured data

Unstructured data such as texts, blogs, documents, photos, videos etc. make up about 95% of stored data (Sarin 2022) and obviously should not be ignored. Machine learning algorithms, some of which are discussed in the previous topic, can unpack and analyse unstructured data in preparation for applying machine learning and visualisation.

When finalising your presentation, it is recommended that checks are performed to ensure key insights obtainable from unstructured data are not overlooked.

Tools to analyse unstructured data include: 

• Apache Hadoop: Comprises a set of tools for data-intensive tasks to clean and extract datasets. MapReduce feature splits large datasets and processes parallel tasks. Apache Pig is a high-level platform for analysing large data sets using MapReduce functionality. 
• Apache Spark supports machine learning, graph processing and streaming computation. 
• Tableau can be integrated with R and Python to prepare unstructured data 
• RapidMiner is an open-source tool for data and text analysis.

Visualisation methods for unstructured data include:

• Word clouds with the frequency of words represented by word size or colour 
• Chord diagrams linking nodes to show relationships
• Heatmaps showing item density, often over a physical location.

Visualising semi-structured data

Semi-structured data is generally easier to deal with than unstructured data, but it still presents challenges. As the volumes of data can be very large, the application of AI machine learning technology is often required.

Analysis tools and formats for unstructured data include: 

• Topic analysis. A machine learning tool scans documents and classifies them based on topic or subject. 
• Sentiment analysis. Automatic analysis of text-based comments (such as feedback or reviews) categorising sentiment as positive, negative or neutral. 
• JSON (a storage format). Using this standard efficient format for data allows for data extraction and analysis with various available tools.
• XML (a simple generalised language). Similar to using JSON for storage, using a common language for analysis is also an advantage. 
• Power BI. This commonly used software has some inbuilt functions for the analysis of unstructured data. Power BI also integrates with other Mircrosoft products such as Azure and Excel, and many third party plug-ins and extensions. 
• Snowflake. Another commonly used software package that can combine, process and analyse different data types.

Visualising structured data

Structured data is generally more straightforward to analyse without the need for preprocessing and applying AI and ML.

This data type has a more standardised and well defined structure; therefore, graphs (line, bar, pie etc) as detailed with examples in the previous topic are most suitable. The types of visualisations include: statistical graphs, plots and infographics.

Commonly used tools for analysing structured data include:

• Microsoft Excel, Power BI, Tableau to produce charts and graphs 
• R Libraries. The open-source R programming language is used by many data scientists and analysts to clean, analyse and graph data. Libraries of tools are freely shared. 
• Shiny application. Shiny is a web interface allowing easy access to tools written in R.

An example of how analysis tools can be applied to unstructured data can be found in this video.

Feedback from stakeholders is essential to the presentation process as it allows for improvements and changes to meet stakeholders expectations. Organisational policies and procedures need to be followed, and they may offer helpful guidance.

It is important to gain genuine feedback from stakeholders professionally and respectfully.

Considerations when seeking and collecting feedback include: 

• Adherence to the organisation’s policies and procedures 
• Keeping feedback and additional comments confidential as required 
• Provision of guidelines on the type of feedback required 
• Guidelines on how to respond to possible critical or negative comments 
• The type of feedback will depend on the questions asked
• Use of open and closed-style questions.

Methods of seeking feedback

Different methods of seeking feedback are available. One or more of these possible options could be utilised:

• Feedback forms. Sometimes this may be handed out as paper-based, but questions can also be asked online.
• Polls or questionnaires. These can web based to collect feedback for specific presentations. 
• Social media. For larger presentations, social media platforms can be used.
• Email. Often less timely feedback can be obtained via email, but this method may allow stakeholders to give a more considered response. 
• Focus group. Feedback can be discussed in person with stakeholders.

Timing of feedback

Audience feedback can be collected during the presentation through polling or questioning, directly afterwards with a questionnaire, or via email or social media later.

Collecting feedback immediately gives more accurate feedback (Collomb 2021). For a presentation, this would mean polling during the presentation or a feedback mechanism directly after the conclusion of the presentation.

Acting on feedback

Integrating feedback into your big data presentation is essential to the improvement process.

After a presentation, it is common for a presenter to rate their own performance, but they will not be able to learn how others perceive them. This is where receiving and acting on feedback is so crucial.

A good presentation relies on collecting in-depth external feedback, carefully considering the feedback and making appropriate changes. The following questions may help decide if corrections are required (Presentationload 2020).

• Is the topic covered in a logical sequence? 
• Do the presentation visualisations clearly show the key messages with an attractive design? 
• Does the presenter use appropriate language and vocabulary for the audience? 
• Is the presenter using appropriate body language for the audience?

The feedback process allows stakeholders to provide opinions and suggestions that the presenter may or may not decide to use to improve their material. It is important to thoroughly analyse received feedback and act on whatever is deemed appropriate. Feedback suggestions are not binding, you may decide to ignore some feedback, but it may be worth recording your reasons. As discussed previously, the original project criteria and organisation practice should be referenced.

A final version of the presentation should be prepared and provided to stakeholders.

Topic summary