#29 Python Interview Questions & Answers | Data Science for Beginners in Tamil

Key Takeaways

Okay. Self evaluation data science list Friday. for a tool exploration. Next week data science question. So in the particular channel Monday to Friday technical topics Saturday one interview so Monday to Sunday to Sunday keep upskilling yourself with a Make a simplified s without any delay. Number one in the interview questions for data science with machine learning course right in the interview questions for data science machine learning course part one it's a checkpoint to validate what you have learned so far So in the conceptual questions and scripting based questions in the conceptual question next week next week Explain fine. Buckle up. Be ready. Okay. So, why is Python preferred in the field of data science? Question answer. Question answer. 70 seconds. 60 to 70 seconds is what something ideal to answer to a particular question. Maximum 90 seconds away. Python is usually having lot of application. Python can be used in web application development. Python can be used in mobile application development and also in data science. So that is why when you compare this Python versus Java or any other programming languages Java, R and this particular programming language is limited to one specific application. R is mostly useful for data analytics. Java is mostly useful for web application development. Python is the only language where it can support uh web application development, game application development, for model building, training, testing, evaluation. We have lot of wide varieties of libraries even in Python. So as Python is having lot of um applications here. So we are preferring Python. And one more important point here is Python is very simple to write. It will be like uh the readability of a particular code it will be easy so that a non-programmer can also understand it. So due to this particular wide varieties of reasons, Python is preferred in the field of data science. What is the difference between Jupyter notebook and Google collab? Jupyter notebook is something uh offline tool. It's one of the multiple ids, integrated development environments through which we can write Python code for data analysis, model machine learning, model building, deep learning and all the kinds of Python based codes we can use in a Jupyter note. The same for Google Collab. The only difference is Google Collab is a cloud-based IDE. Google's cloud-based IDE. The same process whatever we are doing in Jupyter notebook, we can do the same thing in Google Collab. We call it Google Collaboratory. But one point is sometimes the shortcuts and everything this Jupyter notebook will be able to execute it. But in Google collab it takes little bit of time. So that's why mostly the beginners would prefer Jupyter notebooks and even better data analysis steps also we can perform in Jupyter notebook itself. But if someone is not having any kind of uh good high GPU powers or RAM so they can prefer Google collab but they have to compromise on getting some shortcuts like how we easily it is something achievable through Jupyter notebook. When would you use a list over a set or a tpple? Interesting question. To answer this, we know the differences and the properties of list, sets and tpple. When we talk about list, list is something which is which is all heterogeneous elements. It's a one-dimensional u uh it's more than one particular data we can collect. That's why we call this particular we calling this particular concept is coming under Python data structures. List allows duplicate elements. List is mutable. Uh list is indexable and slicable. So if I wanted to gather some values which can be uh which if I wanted to take values inside a list there shouldn't be any restrictions. Later I can also change the values whichever I've updated in a particular list. But when we talk about tpple tpple once the values got assigned to a tpple the values can't be changed in time in the future. So that is why if I if I'm collecting some values list will allow duplicate elements the list values can be changed anytime in the future but tpple will not do that. So whenever I don't want a particular variable not to be changed anytime in the future I will be getting all the particular datas in tpple and other question is sets. Sets is something which uh will not allow duplicate values. So that is the main strength of sets. So when I'm collecting more values no restrictions list. But if I'm collecting multiple values, if duplicate elements has to be eliminated, I'll be collecting in sets. When I'm collecting multiple values, if uh that values once assigned to a variable, if that particular variable if it is not change, if if it should not be changed anytime in the future, that's when I'll be going with uh tpple. Hope it is clear. What is exceptional handling? Give a realtime example of when it's useful. Great. Exceptional handling. Interesting question. So I'll take one example. Realtime example as you've asked me we were doing a project number guessing game. So there from the deployment perspective when a user comes there and there'll be a screen enter your number and user will enter a number between 1 to 100. There are some chances where user will type 50 positive 1 to 100 fine there are some chances where user will enter minus 50. Okay, minus 50 is something an exception, right? And sometimes instead of typing uh 67 or any number, they will type any characters. So this minus 50 or this characters, these are all something which is not in the particular condition which I've already set in the program. So this is a negative uh number error. And the second uh error is out of value error which means the values whatever I'm expecting it is going out like characters. So that is when exceptional handling is nothing but when I'm executing a particular program the program should not crash but the program should say hey these are all the numbers the number what you've entered is not in this particular uh bucket whatever it is mentioned so make sure to write the number properly and at the at the same time if any characters are mentioned it should not throw an error but rather it should handle it properly saying hey this is some characters what you have written so but we were we are expecting some numbers so in this way we will write a try And exceptional block try block will have multiple categories one to can just have multiple conditions like if else uh nested if else nested if so a lot of conditions we can have exceptional block will contain some kind of code through which by mistake if a user is typing any kind of characters this will be able to handle and tell them this is a you are expected to type some kind of values that are numbers not characters so program will execute but it won't crash so in when that particular scenarios I have already used just try exceptional handling. So that's how I used to handle define data frame and series in pandas. What is different data frame and series? Pandas is having multiple data structures. One is uh one-dimensional data structure of pandas we call it a series and uh two-dimensional data structure of pandas we call it as data frame. If I'm getting values in series it will be one dimensional where the series will return index number along with the values. data frame will have both index number row wise and also in column wise it's a two-dimensional uh data that's what we call it as pandas data on one dimensional is serious two dimensions data plane that's the difference what are the different ways of dealing with missing values in the data set this is really an interesting question to deal with missing values firstly I'll be getting the client from the data first there are multiple approaches first I'll just go with step by step number one I'll be asking my client hey the we the data set whatever we have to see it is having some missing values. Uh if you can get the right values to me, I will be able to build a model accordingly. In case if I'm imputing with some kind of other values, I have some some techniques. If I'm doing it, maybe there are chances where the data values it'll be misinterpreted. Um so this is one thing. If you have it, you can give give it to me. Yes, I I'll suggest you some steps. So this is will be my step one. If the client says agrees to it and if he's just giving the real data without missing values, fine. Let's imagine if he's not giving the data and he's asking us to do the perform performing the data cleaning. So that's when I have mean median mode imputation and before that I look into the number of entries. Let's say I have one lakh observations out of which how much percentage of data is uh on this particular record. How much I'll be checking is there any uh drop sorry is there any duplicate values are there or not. First I'll be checking it. I'll be removing all the duplicate entries. there are chances which it'll have and again after dropping it I'll be checking for null null observations if there is any null entries for each and every particular how many particular rows are there I'll compare and I'll again I'll try to remove all this particular null entries wherever the in each if I take any particular record in every record has to be filled so there shouldn't be any record without any values so I'll be removing all this particular stuff so that my data will be clean this is will be my step number two else I'll then again just dig uh dig more and I'll try to observe maybe I'll just particularly categorize in a particular column let's say if I have multiple categories like um level one level two level three and in level one if I have the average salary as 50 50,000 level two uh 53 sorry 60,000 and level three if I have 75,000 and in level one if I have two entries if it is left out I'll try to manage I'll just put a cap fill that particular same the same 50,000 average here it is 60,000 here it is 65,000 So like this there are multiple approaches which I will be following. By looking at the data set I I'll be thinking about it and I'll be coming up with different techniques so that the data will be filled and cleaned properly. How do numpy and pandas differ in terms of functionality. Pandas is a data analysis tool but numpy is a uh uh numerical python tool. So using numpy I will be able to create multiple arrays array multiplication and this uh numpy numerical python is widely used in deep learning applications. But uh pandas is most useful for data analysis stuff. With respect to pandas, I will be able to import a file into my Jupyter notebook and I can perform even data analysis using group of functions. Lot of functions as pandas is having particularly is a data analysis opensourced tool. So that is what the functional differences is. What is the role of library like seaborn mapplot lip in a data project? Great question. Mattplot lip and sebond is one of the um Python based libraries particularly for data visualization. Matt plot lab is something uh it is a data visualization library where the charts and everything whichever it is going to give it won't be something you know it is it won't it would be similar color but uh when I just go for seab bond which is constructed on top of mattplot lib it has some beautifification colors uh it is an kind of add-on or uh it it will have lot of beautifification charts within this seaborn library that is the difference and seaborn won't have much kind of customization option so That's why uh we will be combining seaborn charts with mattplot lip functions to do lot of customizations. So that is why I'll be choosing between the seabond and mattplot. Matt plot lip will more do more customization on the particular chart. Seabond will have more whenever for whatever the charts it is actually trying to show to us. It'll have lot of u um good good presentation or it will be by default it will be colored. That is what the basic difference. When should you use lock and eyelock? Okay, lock is a locator or uh yeah locator. Eyelock is a index locator which means if I wanted to filter a particular column using locator I can use that particular column name or index number. But in Ilocator index locator I will be able to only use that index number of a particular column and index number of any number of rows that you wanted to take. So that's what the main difference locator it will locate based on uh you can give index number or column name both it will accept but in uh Ilocator it will accept only index numbers that's what the difference um so that's what difference and based on however I want to do the filtering I'll be choosing between any of these two what is the importance of writing functions in Python so I will be having step-by-step instructions to perform and uh if the same set of instructions has to be performed again for a particular piece of code for a different input I'll be supplying one input and the same set of instructions has to be performed and when I'm supplying input two if the same set of instructions has to performed again I don't need to write all the codes again so rather I'll I'll bucket all this particular function all the operation step-by-step instructions into a userdefined function and I'll be calling that particular function again and again no matter wherever I'm wherever I'm just calling that particular name inside a program. So to avoid multi writing the same step-by-step instructions again, I will be knocking the door of userdefined functions and uh I'll be calling the particular functions multiple times anywhere in a particular uh program. And um we have uh userdefined function and predefined function. By the way, I've talked about the userdefined function. Predefined function are something the step-by-step instructions are already written. For example, we have a function called len. This len is something which is len. I'll be passing a particular variable or any strings there. It is going to count how many characters in the particular sentence or in a word which is something returning it to me. Right? So it'll perform an operation. It will return something to me whenever I'm just calling that particular function. So this is the answer to the importance of writing functions in Python. Great question. answer it can sound very natural official we can even speak better example this is something. Definitely people are going to be impressed. Okay. Fine. Fine. So in the episode I just tried it. question. So, how do you feel? I would love to read your comments and next week so that you will also understand. I would definitely love to see how do we both match. So I would encourage you all to give some kind of comments and everything. He's a senior product manager. take because so that is what tomorrow's episode is all about and Sweden electrical environmental electrical engineering. It will be definitely motivating for you. So, let's meet in tomorrow's episode. This is your a John. Let's make a simplified. See you all in the next episode. Bye. Cheers. [Music]