Data as evidence for statistical inference
In order to gain understanding of complex processes and systems, we must first gather and analyse data. The variety and amounts of collected and stored data grow every day, boosted by rapid advances of Information Technologies (IT).
Examples are widespread and diverse:
- In the business world, managers use data to analyse the results of marketing studies about new products and to help predict sales.
- In finance, data is used to study stock returns and investment opportunities.
- In health and medicine, data from clinical studies is used to evaluate the efficacy of new drugs and treatments.
However, quantitative data is not information, just ‘numbers’, and it may not be easy to comprehend what those numbers are telling us.
Hence, we need to extract information hidden in the data.
For example, you may want to collect data to make informed conclusions and answer questions of interest, such as:
- How likely are you to win the lottery?
- Is there a bias against women in appointing managers?
- How can you analyse whether a diet works?
Thus, in data science, we face the following fundamental question:
How can we bridge the data and information it bears?
It is important to realise that the data we collect is subject to uncertainty and variability, and the sample we have collected is only a small representation of a larger population behind it.
Statistics provides a powerful and efficient approach to data-related problems as a way of thinking about data and quantifying uncertainty.
Statistics is the art and science of learning from data.
