Understanding Data Science Concepts using Insightful Images

Jai Kushwaha
4 min readOct 6, 2020

Data can be best explained via visualization but can be understood or even remember based on concept/Domain knowledge. However the application of statistics, machine learning and deep learning comes with bucketful of hypothesis, back end algorithms, complex parameters and many postulations.

So to remember all sometimes becomes a burden but human understanding mostly long and short term memory. Long term memory is based on images or glimpse of the past.

Also, human is affected by a selective attention. We see what we want to see sometimes we ignore the behind stories.

Here is a link where your attention mechanism can questioned and answer the following questions

How many passes does the team in white make? Count the number of passes the White team makes. Do you see the Moonwalking bear?

https://www.youtube.com/watch?v=Ahg6qcgoay4&list=PLjvSX3DDjLq36Uy3ZscaSDNd3frGUe8gp&index=2

Here are some images for basic understanding of concepts/notion in the field of Data Science.

Simple Confusion Matrix understanding of FP and FN.

This image is just a means to remember things. As to many concepts becomes cluttered in our mind, so if some ask you type 1 and type 2 error there is the simple example.

So many testing parameters
Basic Confusion Matrix
Definition of Recall , Precision and Accuracy
More detailed definition metrics out of the Confusion Matrix

Now comes the part how we can use these metrics in our testing of training, holdout and test data for effective outcome.

So starting point while we have given data, basic classification on what is to be done with data is learning methods. But here also what to use is the question.

We have the data, then we start with data exploration and many obstacles comes in the way in the form normalization, variable type, missing values, outliers. Here are some illustrations

Gaussian Distribution Telling some statistics
Fitting in the bell curve
  • Fitting in the bell curve
  • How many significant values are there.
  • Problem of Correlation and Causation with respect to continuous and categorical variables respectively.
  • Graph showing different correlation.
Over and Under fitting, significance of data and bias.

Basic Time spend on preparing the data

Justice league and comparison of data science roles and programming languages.

Big data

Customary perception and anticipation across different society segment.

Machine Learning Basic engravings.

Deep Learning Memes

Some funny ones to Conclude

Conclusion:

With reference to data science, modelling, statistics, machine learning and deep learning. There is a enormous pool of data guides, books , courses and many reference available but to grab the concept and application with respect to the business makes the most of this field.

References:

Google Images

Twitter

Pinterest

Instagram

--

--

Jai Kushwaha

I am a 11yrs+ experienced Senior Consultant in Analytics and Model development with domain expertise in BFSI.