Statistics for people in hurry!
Statistics, Being a major subject of mine for consecutive 4 years. I am here to try hard to make a point by proving it simple and quick and basic to understand.
If you want to get familiar with the basic major concepts of statistics in a hurry, then let me summarize it for you.
So Google’s chief Data Analyst was right when she said, ‘Statistics is the science of changing your mind.’
Yeah…as simple as that!
Statistics is a science that changes your mind. As it hits in the face of your prior schemas with new factual realities as it is an aggregate of numerical fact. Statistics deals with what you are not aware of.
By this, we mean that Statistics works on uncertainty. Where there is the probability of occurrence and non-occurrence of possible value, statistics tends to occur there.
Facts are the parameter and not the ultimate guide to statistics. We use samples when we have a deadly demand of Population itself and this reflects uncertainty.
Statistics not only work with data but also makes predictions. The crucial process behind discoveries of science is none other than stats itself.
3 reasons why we need statistics?
1. It provides the proper methods to collect data.
2. It employs the correct analyses.
3. It presents the data effectively.
Descriptive vs. Inferential statistics
Graphs, tables and spreadsheets are the descriptive sides of statistics. Whereas predictions are made under uncertainty and that side is covered under inferential statistics. Inferential statistics allows us to give statements beyond the available data set.
That is why statistics deals with data beyond availability. For instance choosing sample instead of population. Certainly, the possibility of failure is always there in stats.
Catching up on the Hypothesis first.
Why Businesses fail at machine learning?
Hypothesis is tentative statements of what might occur.
Look at this example, We can do lunch together (Default Action) if you usually take 20 minutes to get ready (Null Hypothesis), but if the evidence(Data) suggests it’s longer (Alternative Hypothesis), you can go for lunch by yourself because I’m outta here (Alternative Action).
Data that does not fall within the zone of default action, neglect the null hypothesis. Whereas, the data lying beyond the premises of the Null Hypothesis demanding alternative action tends to justify the Alternative Hypothesis right.
Conventional Statistics vs. Bayesian Statistics
Classical statistics that use least square techniques and maximum likelihood is a conventional type of stats that can be seen in textbooks or academic courses.
In fact, probability calculations belong to Bayesian Statistics. It is a mathematical procedure that applies probability to statistical problems. Moreover, it reports results using credible intervals. Data supposed to lie in between given ranges or intervals is obtained through Bayesian stats.
That’s why Cassie Kozyrkov states that Bayesian statistics updates beliefs by incorporating new data.
Frequency is term that refers mostly occurring data. frequentist statistics is about taking default action. The classical statistics is full of frequentists and you can guess that by entering into your stats101 zone.
When talking about frequencies and classical statistics and uncertainty, Errors evolve overtime. Basically there are two types of error in frequency statistics.
1. Type-1 error (changing your belief when you shouldn’t)
2. Type-2 error (not changing your belief when you should)
The easiest explanation of errors. Type 1 error is like punishing the innocent child and Type 2 error is failing to punish the naughty one. The balance among both errors is amazing unless you obtain more DATA. That’s what statisticians are good at.
Lastly another major concept is Statistical significance. Where the statistics intents to change the perception of people, statistical significance claims that we have changed our perception(mind) in the face of current ground realities.
Summing it all up by giving you a thin line difference between Frequentist and Bayesian statistics that one is about leaving default action while the other one is about having a prior opinion and then changing it with facts based on data.
Author Info:
My name is Akhunzada Younis Said. I am a software project manager in HAZTECH, a software engineering graduate and a content writer. I love working with Linux, Data science and open-source software.
