Blogs

Do you Hate Collaborating?

Written by Abhishek Tiwari

An institution functions smoothly when there is a collaborative effort between multiple stakeholders. School is one of the prime examples of this scenario. One of the best examples of collaborative efforts is a child; both the parents invest their time equally to make a child into a better human being.
Being a science student, I find it suiting to give a scientific example on collaborating. In science, collaborative efforts between several individuals is a key catalyst for great scientific discoveries. One of the most revolutionary biological discoveries – DNA, was discovered by James Watson and Francis Crick. There are several such examples in scientific history; how can we forget Marie and Pierre Curie discovering radium.
... Despite such obvious examples, we tend to be reluctant to work with our peers. One of the reasons that I feel that might be the cause of this behaviour is that we have competitiveness in our nature particularly from our childhood stage when we are taught to compete. This competitiveness leads to a lack of trust between peers and trust is one of the important pillars of collaboration. Although things will change especially once you have an experience of working in a team to achieve a common goal.
The main issue is that in the 21st century, we still need something which will make us understand the importance of teamwork. A recent test match win by India in Australia took multiple efforts from multiple players to cross the line. Mighty West Indies and Great Australian Teams in the ’80s and 2000s were at the top not only because their players were great but because they also knew how to perform as a team.
As a student, it is essential to be a team player which makes student collaboration a necessity. One of the most relatable anecdotes is the group study before the exam or a group project. The beauty of group studies is that even after going to coaching classes, students are able to connect with that subject only through intense discussions with their peers, which might seem disconnected through coaching centre or even through classroom studies.
Now when I think why group study was more successful for me is because of the ease of communication and understanding. My peers who were my friends knew what all examples were needed to help me understand.
I guess in these difficult times we understood the importance of working closely with our peers in a physical environment is very essential for our growth. I really feel that all the students, since the past 10 months missed the experience of physical classrooms which is important for betterment as a human being.
Hence, I suggest to every student that if you want to understand the beauty of collaboration then start participating in hackathons where you give your 48 hours with your team to solve some challenging issues being one of the most effective ways of student collaborations right now.


How to Actually begin your Data Science journey?

Written by Sejal Anand

You may have learnt Python/R for Data Science, studied the mathematics and statistics behind it, understood the steps for implementation of a project but you still can’t figure out how to ‘actually’ start a project.
You are not alone! Every beginner was right there. Keep reading to see a few steps which can take you from ‘a clueless beginner’ and put you right on the first step of your Data Science journey.

Just Start!

...

1.Start a jupyter notebook, pick a project and start writing code.


Pick a beginner-friendly project. Some of the beginner-friendly projects are iris-flower identification, titanic survival prediction, big mart sales prediction among several others. You can find numerous projects including these on Kaggle. Read other people’s code. See how they have approached that specific project. Overall structure and steps to solve projects is similar but if you drill deep down, implementation may vary from person to person. You may not understand what all they have written. But, the most important thing here is to ‘NOT GET INTIMIDATED’. A beginner is not supposed to understand everything. Read the code, search what it means, what it does and in that very process, you are supposed to learn. Now, you might encounter errors in your code and feel stuck. The most simple and convenient solution to the errors when you fail to figure out how to solve it, is to ask in the A CoLab community group which is an ever-growing global community. There are groups on telegram, whatsapp and discord where you can post a query and people of your domain can help you solve your queries.

2.Read


Read blogs and articles relating to projects, approaches, best practices and tips. You will learn a lot in that process. As a beginner, you can look for ‘walk-through’ articles; they will help you simplify your learning process. Reading can also include, reading other people’s jupyter notebooks. After all, jupyter notebooks are not just blocks of codes. Good jupyter notebooks include step-wise explanation and analysis all along.

3.Be Curious


There is no such thing as asking too many questions while solving a data science problem. The more questions you ask, the more you dig deeper into data, drawing patterns, the better the analysis. This will help you learn trends, draw conclusions in a better way and may even help answer questions that you didn’t even know you could ask.
Curiosity will also help expand your knowledge. Asking questions like ‘why’ and ‘how’ will lead you to finding more about that topic, that line of code, that library and will ultimately widen your knowledge bank.

4.Teach


Sounds bizarre? “I am a beginner. How can I teach?”
But it is actually going to make you learn and master your concepts even better. There is a famous quote which says, “When a person teaches, two people learn” and it couldn’t be more right.
You don’t need to teach how to solve an entire project. You must have started small. Pick that small topic, teach your peers and in that process, you are going to research more about it, find a new perspective of questions from your peers and that is going to help you get hold of the concepts in the best way.

At A CoLab, we encourage students to take up SharEd sessions to teach as well as learn from peers. You can become a speaker and along with brushing up your theoretical concepts, you will enhance your communications skills too. In a similar way, you can write blog posts like these too. These will add brownie points to resume as well because if you can explain others well enough, you definitely know it well enough.


How to turn your Jupyter Notebook from “Basic Plain Code’ to ‘Explanatory’?

Written by Sejal Anand

Jupyter notebook is very popular amongst data scientists, and one of the reasons has to be the ability to make jupyter notebooks ‘explanatory’. And we are not talking about just ‘comments’; read further to learn what else you can do with Jupyter notebooks!

...

Does your notebook look like this?

If you said ‘Yes’, let me assure you that you are not alone and probably everyone started out like this.
But, let us learn what all you can incorporate.

1. Comments


This is not a new suggestion if you are a programmer, but it is an important one. Comments in your code will not only help you recall what a particular line of code does or why you imported a package but it will help anyone reading the code, understand it.
Comments don’t need to be long sentences. See how just two words above are enough to explain what the library is for.
Remember, “Any fool can write code that a computer can understand. Good programmers write code that humans can understand.” – Martin Fowler

2. Headings


On the other hand, adding headings will make it easier to figure out what is is about at just one glance.

When you look at the code like the one above it is difficult to make out at once, what all it is about.
It looks a lot more cleaner and well structured.

Also, the code here is fundamental, which may be understood by you even without comments and headings, but as your progress with a project, it does not remain this simple. So, making a habit to put headings will increase the readability of your notebook code.

3. Explain the code


Use the markdown, to write about what you intend to do. For instance, in the image below, specifying that we are looking at the rows which have distance as 0 saves the time and effort of reading the code to figure what is it about.

4. Code output Analysis


Another effective habit to incorporate is to write analysis of the output. This is extremely essential if you are doing data science projects because the main purpose is majorly prediction which involves drawing insights and analysis from data. Ask questions and answer them right with your code.

For instance, in the reference given below, there are rows with distances equal to 0. We try to guess what might be the reasons behind it and write them after the output to make it as easy for the reader of the notebook to understand our analysis as easy it should be for us.

What are the practices you follow to make your notebook to make it explanatory and increase its readability?


Interact Hours #1 “The need to Optimise”

Written by Sejal Anand

InteractHours is an A CoLab’s community groups’ exclusive platform for brainstorming discussions, sharing of resources and knowledge and essentially peer learning.
...

Sakshi Nigam: Hi, I'm Sakshi, pursuing M.Sc. Applied Statistics.
I think optimization of any process helps in increasing the reliability, productivity and efficiency.

Shubhanshi Sharma: To Achieve the best quality with less cost incurred, where possibly the major focus is on productivity efficiency and measurability.

Kounteyo Roy Chowdhury: Hi, I'm Kounteyo from Symbiosis Statistical institute. Recently I used optimization for tuning my language model to get better accuracy.

Aamina: Hey, I'm Aamina, pursuing M.Sc. Applied Statistics and Analytics from NMIMS. Lately, I’ve been reading and researching about Reinforcement Learning and how can we apply it into different sectors to improve performance. I feel it is very dynamic and one of the best methods to solve optimisation problems.

Sarmista: Hi, I am Sarmista pursuing MSc (Data Science). I think optimization is required for dealing with time and space complexity issues of an algorithm. It makes it easy to execute a task.

Aditya: Hi, It's Aditya I am working in NLP based startup from last 2 years.I think optimisation is required for getting the best fit model.

Soham Halbadge: Hello everyone. My name is Soham Halbandge. I'm currently pursuing my bachelor in Statistics. If we are talking about optimization for algorithms. Then I would say optimization is very important for deep learning and other things as well. Training a complex deep learning model can take hours, days, or even weeks. The performance of the optimization algorithm directly affects the model’s training efficiency.
On the other hand, understanding the principles of different optimization algorithms and the role of their parameters will enable us to tune the hyperparameters in a targeted manner to improve the performance of deep learning models.
To train a neural network model, we must define a loss function in order to measure the difference between our model predictions and the label that we want to predict. What we are looking for is a certain set of weights, with which the neural network can make an accurate prediction, which automatically leads to a lower value of the loss function. I think, that the mathematical method behind it is called gradient descent and there are many other optimisation algorithms.

Shaunak Sirodaria: Talking in terms of mathematics, it's basically attaining the zero. Check Numerical Methods.

Sakshi Nigam: In the field finance, optimization is in terms of understanding the risk associated with an individual and providing the bank with a best-fit model to decrease its overall risk.

Sarthak: Hi, I'm Sarthak and I am pursuing M.Sc in Physics. Speaking in terms of physics, optimization is the heart of every possible event in nature. Take for instance the flow of a drop of water on a non-absorbing material (eg the material of raincoats or umbrella). If you have noticed carefully, the path this drop takes exactly mimics the optimization process we involve in deep learning (stochastic gradient descent). So in a way we are trying to follow the laws of nature as far as optimization is concerned.

Sarmista: Reminds me of greedy approach algorithms!

Shaunak Sirodaria: Reminds me of Special Theory of Relativity! Taking the shortest path to reach somewhere, that's nature's fundamental way that runs the universe. Optimization after all? Because you would need more energy to take the longer path, and why put more of it than needed?
So are we trying to mimic that? Yeah now if I see, it's also very inherently natural to optimise.



Hop into our community groups for more such discussions.
(Find the links in the footer)


Interact Hours #2: “How would you explain or teach analytics or data science to a 10th grader?”

Written by Abhishek Tiwari

InteractHours is an A CoLab’s community groups’ exclusive platform for brainstorming discussions, sharing of resources and knowledge and essentially peer learning.
...

Sakshi Nigam: A quick question :
India has reported fresh COVID-19 cases,
Or
India has reported 11,530 fresh COVID-19 cases in the past 10 hours.
Which one is data?

Aditya Hegde: To piggyback on this. I would say
There is information all around you, when you count/measure/assign a number to it, it becomes data.

Sakshi Nigam: Exactly. So anything which is a fact or has Statistics (nothing but numbers or attributes associated with it) is data

Priyank Sahu: Not just numbers; Words collected also comprise data.

Abhishek Tiwari: So any number(real no.) or anything that can be represented as a number is data for me.
When we play gully cricket we keep track of how much run conceded or scored in an over that also comes in the category of data.

Saba Nasir: So in social sciences, data is also qualitative in nature; of course for accurate analysis, one is required to convert it into a quantitative form.
But yes, it is not always numbers.

Shreya: Data is simply information.
And data analysis in the most basic sense is to get some insights about the data at hand.

For example, the marks we get in class. They give us information about the student's knowledge of the subject.

But in order to get an insight or something meaningful out of it: we could probably find averages [mean as you may call it in school/statistical terms]
So we could get insights from the marks data such as "which subject is the class performing the best in?".
We can answer that by taking an average of the marks of all students, subject-wise.
This is what analysing the data is all about: getting meaningful insights using data.
78888 07643: Data science is a broader term that requires knowledge of computer science, statistics and Mathematics.

Bhavya Dharamsey: Data science is basically dealing with data through scientific methods.
Sarmista: The science of getting information from Data is Data Science.
Bhavya Dharamsey: In simple words, Data analytics is understanding and getting insights into the data which we already have.






Hop into our community groups for more such discussions.
(Find the links in the footer)


Want to write a Blog?

Write Here