Every organization today produces data: transactions, sensors, system logs, interactions, questionnaires, images. Collecting it is no longer the point, that happens anyway. The point is knowing what to do with it. This is where data analysis comes in: the discipline that turns raw numbers into strategic decisions.
For a research and development center like PMF Research, which has worked on AR/VR, artificial intelligence (AI), IoT, blockchain, and big data since 2003, data analysis is not an add-on service. It is the layer that holds together almost every innovation project funded by national and European programs.
In this article we look at what data analysis is, how it works, who does it, and why it has become a recurring component of funding calls. Read on to find out more.
Table of contents
What is data analysis?
Data analysis is the process of inspecting, cleaning, transforming, and modeling raw data to produce useful information that supports decision-making. It is an interdisciplinary field that combines statistics, computer science, and business intelligence. In practice, the term is often used interchangeably with data analytics.
The definition, however, only tells half the story. Data analysis does not stop at organizing information: it aims to interpret it, bringing out trends, correlations, and hidden relationships that nobody had looked for before. And it is an iterative process, every cycle of analysis raises new questions, which call for new data and new checks.
A second element separates good analysis from a plain report: the ability to turn a technical result into a recommendation that decision-makers can understand. A flawless statistical model that nobody can read produces no value at all.
How does data analysis work? The 4 stages
The data analysis process is typically structured in four stages.
1. Data collection
You identify the sources like internal databases, IoT sensors, online platforms, open data, questionnaires and acquire the data. Planning is decisive at this point: an unrepresentative sample leads to the wrong conclusions, even with the best analysis further down the line. This stage also has to account for ethical and regulatory constraints, GDPR included.
2. Cleaning and preparation
Real-world data is almost always “dirty”: missing values, inconsistent formats, duplicates, typos. This stage, normalization, variable encoding, feature engineering, aggregation, is by far the most time-consuming, and the old adage holds: garbage in, garbage out. Feed a system flawed data and it will inevitably return wrong or useless results.
3. The analysis itself
This is where the appropriate techniques are applied: statistical tests, machine learning algorithms, text mining, time series analysis. The choice depends on the original question, not on the tool that happens to be available.
4. Validation and communication
Results have to be checked for robustness and generalizability, then delivered through dashboards, visualizations, and reports that stakeholders can actually read.
How many types of data analysis are there?
Data analysis techniques are organized into four levels of increasing maturity, each tied to a different question:
- Descriptive analysis ➝ What happened? It summarizes data using measures of central tendency, frequency distributions, and visualizations. It is the basis of all reporting.
- Diagnostic analysis ➝ Why did it happen? It looks for causes through drill-down, correlation analysis, and benchmarking.
- Predictive analysis ➝ What could happen? It uses historical data, regressions, time series, and machine learning models to estimate future scenarios.
- Prescriptive analysis ➝ What should we do? Optimization, simulation, and decision-support systems suggest the best course of action given the constraints.
Upstream of all of these, exploratory analysis helps you get your bearings in a new dataset and formulate the hypotheses to be tested.
Data analysis tools
There is no universal tool: the choice depends on data volume, the complexity of the analysis, and the skills available in the team. The most widely used are:
- SQL and NoSQL databases, for querying structured and unstructured data;
- Python, with the Pandas, NumPy, SciPy, and scikit-learn ecosystem and the TensorFlow and PyTorch deep learning frameworks;
- R, for advanced statistics and academic research;
- Power BI, Tableau and Looker Studio, for dashboards and data visualization;
- Microsoft Excel, still irreplaceable for quick, small-scale analysis.
In practice, a real project combines several of them, drawing on the strengths of each.
What does a data analyst do?
The data analyst is the professional who oversees the data analysis process. Their core job, in a word, is translation: turning raw data into information that is useful for business or project decisions.
Their typical activities are:
- collecting data and researching sources;
- selecting relevant information and discarding the noise;
- organizing data and building clusters;
- applying statistical methodologies;
- identifying correlations, trends, and recurring patterns;
- creating and updating reporting;
- communicating results to stakeholders.
On the technical side, the role calls for solid foundations in mathematics, statistics, and computer science, command of SQL and of at least one of Python or R, and familiarity with data visualization and business intelligence tools. On the soft skills side, what counts is logical reasoning, problem solving, precision, and above all the ability to communicate.
The data analyst should be distinguished from two adjacent roles. The data scientist starts from the same data but designs mathematical and predictive models, making extensive use of machine learning. The data engineer builds and maintains the infrastructure, pipelines, architectures, data accessibility, that makes the work of the other two possible.
Data analysis in research projects
There is one aspect of data analysis that rarely shows up in general-interest articles, yet is decisive for companies, universities, and public bodies: data analysis has become a cross-cutting work package in almost every funding program.
Horizon Europe, regional ERDF calls, national research and development programs, Erasmus+, funding for the green and digital transition of SMEs are all contexts in which collecting and processing data is not a detail but an evaluation requirement. Calls ask for explicit analysis methodologies, measurable indicators, GDPR-compliant data management plans, validation of results, and dissemination campaigns.
Put plainly: a consortium submitting a project without a data analysis component starts at a serious disadvantage at the evaluation stage. And building that component takes a partner who has worked inside those mechanisms.
Looking for a partner for a research project? You’ve found one
PMF Research is an ICT research and development center based in Catania, Italy, registered in the Innovative SMEs section of the Italian Business Register.
Our R&D activities have received funding from PON Imprese e Competitività (the Italian National Operational Programme for Enterprises and Competitiveness), PON MIUR Ricerca e Competitività 2007–2013, the PO FESR Sicilia and POR Sicilia regional operational programs, Erasmus+, and the Lifelong Learning Programme.
Projects we have delivered include SECESTA ViaSafe and AMELIE, funded under the ERDF Regional Operational Programme Sicily 2014/2020; MINERVA, funded under PON Imprese e Competitività 2014–2020; and Age-SenseAI, a project under Italy’s National Recovery and Resilience Plan (Mission 4 “Education and Research” – Component 2 – Investment 1.3, NextGenerationEU) completed in 2025, in which PMF Research coordinated the work package on data representation and visualization and designed the AWS-based cloud platform where sensor network data and artificial intelligence algorithms come together. Age-SenseAI is a good illustration of what we bring to a consortium: fusing data from heterogeneous sources, processing it with AI techniques, and delivering the results in a readable form.
If you are putting together a partnership for a national or European project involving data collection, processing, or analysis, we can work alongside you as a technical or scientific partner, from defining the methodological approach to designing the data architecture, from developing the analytical models to validating the results and communicating them.
Let’s talk. Get in touch with a short description of your proposal, and we will come back to you with a concrete assessment of how we could work together.
