Data Analytics

Turning a vague question into one data can answer.

• The question behind the question
  "Sales are down" is not an analysis brief. Getting to "which branch, which product, since when, compared with what" before touching a spreadsheet.
• Metrics that mean something
  Choosing a measure a decision can act on, and spotting the vanity number that only ever goes up.
• Where the data lives hands on
  Point-of-sale exports, an ERP, a Google Sheet, a WhatsApp group. Real Pakistani businesses, realistically.
• How analysis misleads honestly
  Survivorship, selection, and the average that hides two different populations.

The tool most of your clients will actually have.

• Cleaning data that arrived badly hands on
  Inconsistent phone formats, merged cells, dates as text, trailing spaces. Where most of the job's hours really go.
• Lookups, pivots and dynamic ranges hands on
  XLOOKUP, pivot tables, and building something that does not break when a row is added.
• Power Query for repeatable cleaning hands on
  Doing the cleaning once so next month is a refresh, not a repeat.
• A workbook someone else can use hands on
  Documented, protected, and not dependent on you being in the room.

Getting the data yourself instead of asking for an export.

• SELECT, WHERE, ORDER BY, LIMIT hands on
  Reading data with intent.
• JOINs, and what happens when they go wrong hands on
  Inner versus left, and the duplicate-row explosion that quietly doubles a revenue figure.
• GROUP BY and aggregation hands on
  Counts, sums, averages, and HAVING versus WHERE.
• Window functions hands on
  Running totals, rankings and month-on-month change — where SQL stops being a data-fetching tool and becomes an analysis one.

Power BI and Looker Studio, and the judgement behind them.

• Choosing the right chart hands on
  Why a pie chart with nine slices communicates nothing, and what to use instead.
• Building in Power BI hands on
  Data model, relationships, measures and a dashboard that loads quickly.
• Looker Studio for clients with no licence hands on
  The free route, and where it runs out.
• Designing for the person who will look at it once a week
  The single number at the top, and what belongs behind a click.

Enough to avoid being confidently wrong.

• Distributions, and why the average is often the wrong summary
  Median, spread and the long tail — with real salary and sales data.
• Correlation is not causation, demonstrated
  Working through a real pair of correlated series that have nothing to do with each other.
• Is this difference real? hands on
  Sample size, variation and the practical version of significance — enough to know when a 3% lift is noise.
• Forecasting simply, and its limits
  Trend and seasonality, and saying plainly how uncertain a forecast is.

The part that decides whether any of it gets used.

• Structuring a findings report hands on
  Answer first, evidence second, method last — the opposite of how the work was done.
• Presenting to people who do not like numbers hands on
  Rehearsing the one sentence a manager will repeat to their boss.
• Saying "the data does not show that"
  Holding a finding under pressure, and how to disagree with a client without losing them.
• Documenting so the analysis can be repeated
  Sources, assumptions and steps — the difference between an analysis and a one-off spreadsheet.

Pakistan's National Vocational & Technical Training Commission publishes a course-contents document for the trades it funds. This programme is mapped to Big Data Analytics — 12 weeks, 4 hours a day, 240 contact hours.

Published by NAVTTC and reproduced here so you can check our syllabus against it. We teach it to those hours: 240 contact hours over 12 weeks, the same as the standard. Read the NAVTTC document.

Before you start

• Comfortable with a computer and basic arithmetic
  No statistics background required; the maths is taught from the ground up.
• Some spreadsheet experience helps
  Not required, but students who have never used Excel find the first fortnight heavy.

Where it leads

• Machine Learning
  The natural next step for those who want to predict rather than explain.
• Python Programming
  Where analysis stops being manual and starts being automated.

Roles this prepares you for

• Data Analyst
  In-house at a business, or with an agency.
• MIS / Reporting Officer
  A very common role in Pakistani firms, and the one this course most directly prepares for.
• Business Intelligence Associate
  Dashboard and reporting work.

Interested in this programme?

Try the subject free first — there is no reason to pay before you know you like it.