Last Updated on August 20, 2022
Content Overview
Statistics are like bikinis. What they reveal is suggestive, but what they conceal is vital.
-Aaron Levenstein, Business Professor at Baruch College
What is Statistics Good For?
Statistics is the discipline of analyzing data. As such it intersects heavily with data science, machine learning and,
of course, traditional statistical analysis. In this post, I orient you to statistics by covering a few key activities that define the analytics field. Simplistically, analytics can be divided into four key categories. These are:
• Descriptive statistics – (EDA, quantification, summarization, clustering)
• Inference – (estimation, sampling, variability, defining populations)
• Prediction- (machine learning, supervised learning)
• Experimental Design – (the process of designing experiments)
I’ll explain these four in more detail below.
Descriptive Statistics – What is happening?
This is the most common of all forms. In business it provides the analyst a view of key metrics and measures within the business. Descriptive statistics includes exploratory data analysis, unsupervised learning, clustering and basic data summaries. Descriptive statistics have many uses, most notably helping us get familiar with a data set. Descriptive statistics usually are the starting point for any analysis. Often, descriptive statistics help us arrive at hypotheses to be tested later with more formal inference.
Descriptive statistics are very important because if we simply presented our raw data it would be hard to visulize what the data was showing, especially if there was a lot of it. Descriptive statistics therefore enables us to present the data in a more meaningful way, which allows simpler interpretation of the data. For example, if we had the results of 1000 students’ marks for a particular for SAT exam, we may be interested in the overall performance of those students. We would also be interested in the distribution or spread of the marks. Descriptive statistics allow us to do this.
Lets take an another example like an data analyst could have data on a large population of customers.Understanding demographic information on their customers (e.g. 20% of our customers are self-employed) would be categorized as “descriptive analytics”. Utilizing effective visualization tools enhances the message of descriptive analytics.
Inference statistics
Statistical inference – the process of making conclusions about populations from a sample
Inference is the process of making conclusions about populations from samples. Inference includes most of the activities traditionally associated with statistics such as: estimation, confidence intervals, hypothesis tests and variability. Inference forces us to formally define targets of estimations or hypotheses. It forces us to think about the population that we’re trying to generalize to from our sample.
Prediction – What is likely to happen?
Prediction overlaps quite a bit with inference, but modern prediction tends to have a different mindset. Prediction is the process of trying to guess an outcome given a set of realizations of the outcome and some predictors. Machine
learning, regression, deep learning, boosting, random forests and logistic regression are all prediction algorithms.
If the target of prediction is binary or categorical, prediction is often called classification. In modern prediction, emphasis shifts from building small, parsimonious, interpretable models to focusing on prediction performance, often estimated via cross validation. Generalizability is often given not by a sampling model, as in traditional inference, but by challenging the algorithm on novel datasets. Prediction has transformed many fields include e-commerce, marketing and financial forecasting.
Experimental Design
Experimental design is the act of controlling your experimental process to optimize the chance of arriving at sound
conclusions. The most notable example of experimental design is randomization. In randomization a treatment is
randomized across experimental units to make treatment groups as comparable as possible. Clinical trials and A/B
testing both employ randomization. In random sampling, one tries to randomly sample from a population of interest
to get better generalizability of the results to the population. Many election polls try to get a random sample.