Data Analytics Fundamentals
What this domain covers
You need a working vocabulary for the analytics lifecycle: problem definition, data collection, cleaning, exploration, modelling, interpretation, and communication. Know the difference between descriptive (what happened), diagnostic (why it happened), predictive (what is likely to happen), and prescriptive (what should be done) analytics. Understand data types: qualitative vs quantitative, nominal, ordinal, interval, and ratio scales, and discrete vs continuous measures. Distinguish structured data (tables, relational databases) from semi-structured (JSON, XML) and unstructured (text, images, audio).
Concepts to master
Be clear on population vs sample, and on the meaning of a variable, observation, and record. Know common summary statistics (mean, median, mode, range, variance, standard deviation) and when each is appropriate. Understand data quality dimensions: accuracy, completeness, consistency, timeliness, validity, and uniqueness. Recognise the purpose of metadata and data dictionaries.
Traps in questions
- Confusing ordinal with interval: rankings are ordinal; temperatures on Celsius are interval because zero is arbitrary.
- Assuming the mean is always the best average: skewed data is often better summarised by the median.
- Mixing up correlation with causation, or treating a sample result as certainty about the population.
- Calling a spreadsheet column of postcodes numeric when it is really a nominal identifier.
Worked rule of thumb
Ask two questions of any dataset: what does one row represent? and what scale is each column on? For example, a customer table where each row is one customer, with columns for age (ratio), satisfaction rating 1-5 (ordinal), and city (nominal), tells you immediately that averaging the rating is defensible only with caution and averaging the city is meaningless.