Machine Learning is Fun – Part 6 – Types of Statistics

Let’s start today’s topic – Types of Statistics.

1. Descriptive statistics – Suppose you have some set of samples and you need to do some kind of analysis with the data.

Before directly jumping into analysis first of all you need to arrange the data, format the data and try to get the basic structure of the sample data.

Here the central of tendency and variance come into play for a major role.

We need to find out the answers to the below questions:

Is my data properly arranged?

What is the behaviour of the data?

How to get the central tendency and variance, and then need to understand the pattern of data.

Is there any correlation among variables?

Any odd input present or not?

What kind of variables are present in the sample?

What is the requirement with the data?

All this is called EDA ( Exploratory Data Analysis).

2. Inferential statistics – After getting all the above questions , you need to sit back and relax with a cup of tea.☕

Because you have done 70℅  of your analysis ??.

So what is the rest 30%?

The most interesting task in my life ever, and that is choosing the model and validation.

Wait.. it’s not easy though you have performed  the major tasks on first part of EDA.

To do the tasks smoothly  in  part 2 , you need to learn lots of algorithm and you must have to clear on requirement properly.

Let me mention some parameters which play major role to choose model after multiple trials.

1. Residuals.

2. Correlation Coefficient.

3. Coefficient of Determination.

4. AIC, BIC.

5. Confidence Interval.

6. Standard error, Marginal error.

7. P-stat, T-stat, Z-stat, F-stat.

8. Chi sq test value.

9. ANOVA.

10. ANCOVA.

Well, that’s it for today. I shall discuss on the algorithms which helps to choose good models for inferential statistics in my next post.

Stay tuned !☺