Data science vs data analysis: the difference, the jobs and the certificates

Data analysis explains what happened and why, using spreadsheets, SQL, statistics and dashboards; data science builds models that predict or recommend, using programming and machine learning. SDAIA's 2025 occupational framework lists them as separate jobs, and its data scientist card expects machine-learning experience and usually a master's or doctorate. Most companies need an analyst first.

This guide compares data science vs data analysis for two readers: owners and managers deciding whom to hire, and job seekers in Saudi Arabia choosing a path. It draws on the National Occupational Standard Framework for Data and Artificial Intelligence that the Saudi Data and AI Authority (SDAIA) published in 2025 [1], and on certificate details read on the issuers' own pages on 14 October 2026. O AI is not affiliated with Microsoft, Google, IBM, Coursera or any other company named here.

Data science vs data analysis: the difference in one table

The two fields overlap, so definitions help. The US National Institute of Standards and Technology (NIST) defines analytics as the systematic processing of data to uncover patterns, relationships and historical trends, and to attempt predictions. It defines data science as building useful knowledge directly from data, through discovery or through forming and testing hypotheses [2]. SDAIA's job cards draw the practical line: the data analyst collects, organises and studies data to give the business insight, while the data scientist uses machine learning and predictive modelling to spot trends, extract information from unstructured sources and provide automated recommendations [1].

Sources for the table: SDAIA's job cards [1] and Microsoft's role descriptions [3].

The two jobs feed each other. In Microsoft's worked scenario, a data scientist trains a revenue forecast on three years of sales, and the data analyst then shows those predictions in a sales executive's dashboard [3]. In Microsoft's own terms, descriptive and diagnostic analytics answer "what happened?" and "why?", while predictive and prescriptive analytics estimate what comes next and compare options [4].

What does a data analyst do? The tasks on SDAIA's job card

SDAIA's card for the data analyst, occupation code D-An, lists twelve key tasks [1]. In plain terms, an analyst:

• identifies, collects and prepares data from different sources so it is ready to analyse;
• checks that source data, and the findings drawn from it, are valid;
• collects metrics and trend data;
• runs exploratory analysis and statistical tests to inspect, explore and visualise data;
• applies statistical techniques to find relationships that can improve processes and support decisions;
• turns all of this into actionable recommendations, and supports data governance and quality teams in setting standards.

The skills on the same card include writing queries and reports, regression analysis, descriptive statistics and identifying outliers, and the abilities include producing clear charts and using data visualisation tools [1].

Microsoft's description of the role adds the day-to-day reality. The analyst builds the dashboard a manager uses, chooses the right visuals, sets up row-level security so regional managers see only their own region, and schedules the daily refresh. The value of the job, in Microsoft's framing, is making sure the output answers what the business actually needs to know [3].

A typical week for an analyst in a Saudi company

Illustrative example. Sunday: refresh the weekly sales report and reconcile its total with the accounting system. Monday and Tuesday: investigate why one branch's returns rose. Wednesday: clean a new export from the online store whose product codes do not match the point-of-sale system. Thursday: present two charts and one recommendation to the operations manager.

What does a data scientist do?

SDAIA's card for the data scientist, code D-S, centres on models [1]. The key tasks include solving problems with applied mathematics, statistics and probability; building and evaluating statistical and machine-learning models that produce insights and predictions; testing model accuracy with performance metrics and hypothesis tests; building reusable data-cleaning processes; and working with machine-learning engineers to deploy models for business uses such as recommendation systems, fraud detection and demand forecasting [1].

The card's requirements are heavier than the analyst's. Alongside a bachelor's degree, it lists a master's or doctorate in machine learning, data science or a related quantitative field, plus experience in programming and in statistical modelling or machine learning [1]. NIST describes the data scientist as someone who combines business needs, domain knowledge, analytical skills and software engineering across the whole analytics life cycle [2].

That breadth is why one data scientist rarely succeeds alone in a company whose data is still scattered across spreadsheets. A model is only as reliable as the data and the pipelines behind it.

The wider data team: engineers, BI and governance roles

SDAIA's framework groups 16 data and AI occupations into five categories: business and leadership; governance, management and risk analysis; data science and analytics; engineering and architecture; and research [1]. The data analyst sits under the specialty of data analytics and business intelligence, and the data scientist under data science, both in the data science and analytics category [1].

Sources for the table: SDAIA's framework [1].

Microsoft adds a bridging role, the analytics engineer, who settles definitions such as "net revenue" in one certified model so every report uses the same figure [3]. Saudi Arabia's National Data Management Office (NDMO) expects a data analytics team to typically include data scientists, data engineers and visualisation engineers [5]. Those standards are written for public entities and business partners handling government data, but the team shape applies anywhere.

Which should a company hire first?

For most Saudi small and mid-sized companies, the honest answer is a data analyst, or an existing employee trained as one. Data science pays off only after the basic reporting is reliable.

If the data sits in disconnected systems, our guide to custom software development in Saudi Arabia explains when to connect what you have rather than buy something new.

Illustrative example: is a forecast worth a data scientist?

Illustrative example. A food distributor in Jeddah estimates that stock-outs cost it about SAR 40,000 a month in lost gross margin. If a demand forecast halved them, it would protect about SAR 20,000 a month, or SAR 240,000 a year. That is the ceiling to compare against the full yearly cost of a data scientist, tools and the data engineering the model needs.

Before hiring, write the use case down the way NDMO asks public entities to: the objective, the type of analytics, the expected benefit, the owner and users, the data sources and fields, and the technology required [5]. If the distributor cannot list two years of clean sales by product and branch, its first hire is an analyst who can produce that history.

Data analysis certificates and data science certificates

Certificates show that someone has covered a defined syllabus; employers still test what candidates can do with real data. The table lists four well-known options as their official pages showed them on 14 October 2026.

The retired DP-100 entry is a reminder to check the issuer's page before you pay for a course built around an exam. Older Arabic articles still recommend it.

What a data analysis certificate proves, and what it does not

A data analysis certificate shows that you have studied the syllabus and, for proctored exams such as PL-300, passed a timed test. It does not show that you can clean a messy export from a Saudi point-of-sale system or explain a result to a manager. Pair every certificate with two or three portfolio projects.

Does HRDF support data certificates?

The Human Resources Development Fund (HRDF) reimburses the cost of approved professional certificates for Saudi citizens, whether employed in the public or private sector or looking for work [10]. Its published conditions, as of 14 October 2026:

• the certificate must be on HRDF's approved list, and obtained after the date it was approved;
• the applicant declares that the employer did not pay for it;
• no more than two certificates per person;
• the certificate must not be more than six months past its issue date when you apply;
• membership fees are not covered, and reimbursement follows a fixed amount per certificate, with training costs capped;
• eligibility review and payment take 40 working days from a complete application [10].

HRDF states that its list changes with labour-market needs [10], so check that the exact certificate is on it before you enrol.

A learning path for job seekers in Saudi Arabia

Use SDAIA's job cards as your checklist: they list the tasks, skills and knowledge for each role and are meant to be used for job descriptions, hiring and career planning [1]. A practical order:

1. Spreadsheets: pivot tables, lookups, charts and cleaning.
2. SQL: querying and joining tables.
3. A BI tool: Power BI or a similar tool, including a data model and a shared report.
4. Statistics: descriptive statistics, sampling, tests and regression.
5. Python: data cleaning with pandas and simple automation.
6. Machine learning: only if you are heading for data science, after the first five are solid.

Build three portfolio projects, each with a question, the cleaning you did, the result and the recommendation. Use public or synthetic data, never a former employer's customer records. Under the PDPL's Implementing Regulation, data counts as anonymised, and no longer personal, only when identifiers are removed so that re-identifying the person is impossible [11]. Deleting the name column alone is not enough.

Your next step

If you run a company, write one use-case card for the question that costs you most, list the data it needs, and let that card decide whether your first hire in the data science vs data analysis choice is an analyst, an engineer or a scientist. For the wider picture of AI readiness, our guide to SDAIA's AI adoption framework shows what a smaller company can borrow from it. If you are a job seeker, open SDAIA's job card for the role you want, mark the skills you can already prove, and pick one certificate that covers the biggest gap. You can read about O AI, a software and AI company in Al Khobar, on our home page.