Skip to main content
Institute of Advanced Technological and Commercial Studies

Skills

Preparing and validating a dataset

An analysis is worth exactly what its input data is worth. Cleaning, reconciling, deduplicating, documenting: this is the work that takes up most of a data professional's time, and it decides how much trust every downstream figure earns.

The work that makes a figure defensible

Two departments announcing two different figures for the same reality: the situation is commonplace, and it almost always has the same cause — implicit preparation rules known to one person only. The skill consists in making those rules explicit: which source prevails, how two identifiers are matched, what happens to a missing value, when a row is rejected. Once written, the rules can be checked, replayed and handed on — and the figure becomes defensible in a meeting.

  • Set the cleaning rules for a source and make them reproducible
  • Reconcile two reference systems that use different identifiers
  • Put automatic quality checks on an incoming data flow
  • Document the provenance and definition of every indicator produced

The blocks that build it

The skill is built block after block, in the order of the years. You see here exactly where it is taught, and for how many taught hours.

Teaching blockProgrammeTaught hoursYear
Data quality and preparationBusiness Intelligence and Steering200 hYear 1
Data engineeringData Science programme — Bac+5 level240 hYear 1
Cleaning and preparing dataCertificate in Artificial Intelligence for Business Decisions40 hYear 1
Databases and queryingBusiness Intelligence and Steering240 hYear 1

Which positions it leads to

At levels 6 and 7, professional experience is part of the entry requirements.

Professional qualificationTarget occupationLevelTaught volume
Data and Annotation Operatordata operator, annotator, dataset preparer4 · Skilled employee and operative600 h
Data and Applied AI Techniciandata technician, AI solutions integrator5 · Senior technician and supervisor700 h
Data Analystdata analyst6 · Manager and professional750 h

The level you reach

You take on a raw source and deliver a controlled, documented dataset that replays identically. Level shows in volumes as much as in titles: here are both, side by side.

teaching blocks
4

Catalogue rule: one credit is twenty-five hours of work, ten of them taught.

of taught instruction
720 h
credits in total
72
professional qualifications concerned
3

The available entry points

The level shown indicates the award targeted; the duration, the real commitment in months of taught work.

ProgrammeLevelDurationFees
Business Intelligence and SteeringBachelor's degree2 years5 000 000 GNF per year
Data Science programme — Bac+5 levelMaster's degree2 years7 200 000 GNF per year
Certificate in Artificial Intelligence for Business DecisionsProfessional certification5 months3 600 000 GNF in total

Before you decide

Is this skill relevant to someone outside computing?

It is first of all relevant to those who produce figures: management control, operations, human resources, compliance. The four-block certificate was designed for them.

What tooling supports the coursework?

Spreadsheets for everyday volumes, SQL and Python for larger ones. The reasoning taught applies to all three, which makes the choice of tool secondary.

How much of a real project does preparation take?

Most of the schedule, and the programme reflects that: 200 taught hours go to it, against 160 for modelling. That is a deliberate curriculum choice.

Preparing and validating a dataset — Skills | IHETC — IHETC