What is data science?
Table of Contents Show Hide
Definition of data science
From Wikipedia
Data science is a multi-disciplinary field that uses scientific methods, processes, algorithms and systems to extract knowledge and insights from structured and unstructured data. Data science is related to data mining and big data.
Data science is broader than any single technique — it spans statistics, machine learning, optimisation algorithms, and increasingly large language models, applied to real business data to answer questions or automate decisions that would otherwise require manual analysis. It’s distinct from business intelligence (which primarily reports on what has already happened) in that data science is often used to predict what’s likely to happen next, or to find patterns in data too large or too complex for a person to spot by inspection.
Executive view
Data science applies scientific methods to data to reveal insights and predictions that can help you to meet your business goals. There are two key goals in applying data science to your business: building better products and making better decisions. Data science is applied to building intelligent features and automation into products to make them more appealing and profitable to customers. Data analysis and machine learning can be applied to business metrics to drive better recommendations and predictions.
Data science helps businesses:
build better products.
make better decisions.
Business function leader view
Data science can be applied in all areas of your business. It helps development teams to improve the products they build, helps sales and marketing teams with market segmentation and advertising, and supports financial decision-making.
You may need this service if:
you are building a product that could be improved by intelligent features.
your team has access to large volumes of data but is not using it to make predictions or decisions.
KPIs you should consider measuring for this are:
increased sales of your product
improved customer retention
increased lead generation
improved cash flow
increased profits
Technical view
Data science can be applied to your product by integrating usage data collection solutions and implementing features that analyse the usage data and apply algorithms to the usage data to drive improved performance and user experience.
Data science helps deliver:
improved performance through analysis of usage data.
personalised content that can drive increased up-sells and interactions.
Get this service if you encounter:
missed opportunities to apply usage data to improving performance.
lack of personalised content in your product.
lack of understanding of how your product can be improved.
Key criteria to consider are:
Do you have appropriate data storage solutions in place that will keep user data secure?
Do you have the time and resources to apply your analysis to make improvements to your product?
Does your product have enough users to generate actionable data insights?
Data science in metals manufacturing
Two applications of data science are especially concrete in steel and metals manufacturing, because they replace work that’s currently done by intuition and a spreadsheet rather than a systematic method.
Cutting optimisation is a classic combinatorial optimisation problem — the cutting stock problem, well studied in operations research. Given a set of available stock lengths and a set of required cut lengths, the goal is to find the combination that minimises waste. This isn’t a simple problem: the number of possible combinations grows exponentially as the number of order lines and stock lengths increase, well beyond what a person working from experience can evaluate by hand. Genetic algorithms and other heuristic solvers — a data science technique, distinct from generative AI — can search that combinatorial space systematically and typically reduce scrap by several percentage points compared to manual planning.
Document data extraction — reading a mill certificate and turning it into structured fields (heat number, chemical composition, mechanical properties) — is a machine learning and AI problem, using document intelligence and layout-aware extraction models rather than a person manually re-typing values from a PDF. Every mill formats its certificates slightly differently, which is exactly the kind of pattern-recognition-across-variation problem that data science techniques are suited to and manual, rule-based approaches struggle with.
Beyond these two, manufacturers are increasingly applying data science to demand forecasting (predicting order volume by product and grade to inform purchasing) and predictive scrap analysis (identifying which order and material combinations are likely to produce higher-than-average waste before the cut is made).