What is logarithm? | Math, Statistics for data science, machine learning

[@codebasics]

Adding little clarification here: in the beginning of the presentation I meant to have 1$ starting amount which becomes 5$ after one year, 25 after 2 years and 125$ after 3 years. When we use log 125 base 5, the base 5 indicates the 5x returns that you get every Year

[@timeforsometruth4512]

Aaah thank you, that answers my question thanks again.

[@AmiteshKeSath]

Got you Sir 👍

[@kunalpandya8468]

Can we use normalisation instead of logarithmic transformation?

[@bunny1240]

If possible could you please share why is natural log of negative number invalid?

[@mjshaheed]

Thank you, Sir!

[@niluthpalchowdhury]

If I had teacher like you in my high school I would have never got afraid of Maths

[@GZAMORA27]

Thanks a lot! This not only provide better insights of the usage of this topics in ML but also help to see what needed to be studied to become a better data scientist

[@codebasics]

Glad it was helpful!

[@kyawswarthant708]

short to the points and easy to understand wish I should come to this channel earlier before going heavy math courses related to machine learning

[@That_Coding_Channel]

This is one of the to-the-point explanations I learnt so far. Thanks a lot!

[@codebasics]

Glad it was helpful!

[@sarthaksanatani05]

I didnt know log concept was so easy . I thought it was fully abstract and need to accept as it is since school and college lectures. Lol

[@adaariina7175]

Wao! I understand you better then my teacher back in school. So easy to understand.

[@sahilshahane1547]

Thank you, now i have more clear understanding of Log in ML's prespective

[@sony7781]

can't wait for more videos. you make the topic so easy to understand with perfect examples. please upload longer videos covering more topics. Looking forward to learn probability and statistics from you which i find difficult with few topics.

[@codebasics]

Next video will premier tomorrow :)

[@sarahlatifkhan8634]

Thank you for the best explanation of everything, I enjoyed your DSAlgo series now revising these concepts, you are so good at teaching! Prayers & Best regards!

[@codebasics]

Glad it was helpful!

[@supersql8406]

Damn! I just subscribed to this guy like a min ago watching his Naive Bayes video and now I fully understand Logarithm!

[@Lorena-qp2ou]

this is the best youtube channel ever

[@vikashdas1852]

Currently I'm learning ML from this channel and I had just started wondering that a resource is required for a grip on fundamentals of maths and boom....this guys again comes to rescue

[@codebasics]

I am happy this was helpful to you.

[@ramesha8539]

Short and crisp videos amazing superb skills 👌

[@gustavojuantorena]

This is an awesome approach to learn math for DS and ML, It's practical and well explained with examples in code.

[@codebasics]

Glad you liked it

[@KumR]

Wish I had someone like you during my college days to teach logs

[@arunavrath8400]

Superbly explained. Thanks Dhaval !

[@giovanniberardi4134]

Very good. Thanks 👌

[@codebasics]

Glad it was helpful!

[@dibyarupnath]

This playlist is very very helpful. Please make more videos for this sIr. Great Work!

[@saikatdutta1991]

I appreciate your diligence

[@azrflourish9032]

THIS IS What I was looking for. Simple explanation!!

[@codebasics]

Glad it was helpful!

[@sejalanand23]

Really excited about this series

[@codebasics]

Keep watching

[@timeforsometruth4512]

@@codebasics hi sir, just a quick question, at the start of the video it says it will take us 2 years to get 5 dollars to 125 dollars at 5x return, as it will be 25 dollars after 1 year and then 125 dollars after 2...why does the answer become 3 years when performing the log function? Wouldn't that be like starting with 1 dollar, then having 5 after year 1 at 5x return, then 25 dollars after 2 years at 5x return and then 125 dollars after 3 years? Please help me understanding where I am going wrong with my thinking. Thank you kindly for the series again.

[@balajikartheek285]

@@timeforsometruth4512 exactly, me too having the same doubt???

[@timeforsometruth4512]

@@balajikartheek285 I see that we were right Balaji. Author has clarified that he did mean to start with 1 dollar, and not 5 👍

[@vivekpaliwal1876]

Best explanation...i will never forget this

[@learnerlearner4090]

Really good info.. I learnt the basics of logarithms and how it can be used in DS. I don't in school I even understood why it's used. Thanks so much for sharing the info.!

[@codebasics]

Glad it was helpful!

[@gouravsapra8668]

Very well explained....Thank you 👍

[@ArunKumar_237]

Thank you guruji 🙏. Please continue this series

[@codebasics]

Keep watching

[@RFICick22]

Excellent, very accurate explaining about basics of log and their use in ML.

[@codebasics]

Glad it was helpful!

[@faith3220a]

These are awesome and very simple to understand... Thank you

[@haripriyamoorthy5290]

Waiting 😎

[@baskarangopalan9291]

Very nice helped me in machine learning terms and I can use this video for my daughter to understand log is inverse of exponents

[@codebasics]

Glad it was helpful!

[@satya6804]

Thanks for great effort in explaining and building this series. please keep it up

[@codebasics]

Glad it was helpful!

[@flamboyantperson5936]

Hey Dhaval, what a great and simple explanation :)

[@codebasics]

Glad it was helpful!

[@ketanbutte3497]

simple, easy to understand, great work

[@codebasics]

Glad it helped!

[@adityabhadoriya3191]

waiting for your next video

[@codebasics]

Keep watching

[@TheMISBlog]

Great Video, Very useful, Thanks

[@codebasics]

Glad it was helpful!

[@instantlearning3003]

thank you so much sir for this great playlist

[@codebasics]

Glad it was helpful!

[@AmarKumar-rp4uw]

Great sir...

[@codebasics]

Glad it was helpful!

[@saurabhbarasiya4721]

Great video sir

[@codebasics]

Glad it was helpful!

[@muhammadnauman1454]

Amazing! i had many charts in my reports like that are difficult to compare so after watching this i added log column and created new charts based on log column now its comparable .

[@codebasics]

Glad it was helpful!

[@MuhammadUmar-px6ij]

Great example

[@codebasics]

I am glad you liked it

[@denuwanharshana4872]

Thank you Sir, this is very usefull

[@codebasics]

Glad it was helpful!

[@Mano64902]

Great explanation

[@Hengul]

That was really really awesome Sir... Got the concepts very clearly.. ❤️

[@codebasics]

Glad it was helpful!

[@deeptidas7246]

Great explanation. Thanks 😊

[@dra.talwar4592]

Its a good idea, clearing concept with practice on Jupyter notebook

[@rabiafyz9952]

Please show some projects of data science that are used in a job or in work. Thanks.

[@AmiteshKeSath]

Sir you are simply awesome... Its so easy and interesting to understand the concept.. million thanks for making such videos..please please please keep doing such great work..❤️❤️🎉🙏😊🙏🙏🙏🌈🌻

[@codebasics]

It's my pleasure, Glad it was helpful!

[@alifaisal253]

amazingly explained

[@codebasics]

Glad it was helpful!

[@mohs8789]

Great ....Nicely explained 👍

[@codebasics]

Thank you 🙂

[@MLDSInsights]

Awesome video sir. If possible please add little more in depth concept and calculation and try adding numerical mathematic problems as exercise.

[@1234569312]

Hi this is a good initiative but please also try to explain the maths behind and try to explain more....it was too short .. Keep it up!!

[@vikashdas1852]

Saying Thank You wont be enough!!

[@codebasics]

Glad it was helpful!

[@asimshah2889]

great explanation sir very helpful

[@codebasics]

I am happy this was helpful to you.

[@chatchaikomrangded960]

Subscribed, I like the way you explain.

[@codebasics]

Glad it was helpful!

[@DibyenduFX]

Very helpful

[@harshadasuryawanshi796]

I woulde suggest Normalization as another topic which is must in this series

[@piyalikarmakar5979]

Sir, it will be helpful if you kindly do a vedio on bias and variance, likelihood Estimations

[@thejillykilly]

Video on loss function please

[@zergham001]

Thank you ❤️

[@hemanthkumar1466]

sir i would like to learn data science and i had a knowledge in python so should i learn libraries first or mathematics topics first for data science from your play list thank you.

[@lol-ki5pd]

I have a question, why did not we treat pooja value as an outlier?

[@Giratina155]

Thank you!

[@programmingwithjackchew903]

what is difference between minmaxscale and log for transforming to a similar scale?

[@Rumil_]

Wow thanks so much! What is the intuition around using log base 10 on poojas number and not some other base?

[@AshishPatel-kn3kc]

Is there any video on R2 ?

[@pankajjoshi8292]

Sir @5:45 why Log base 10 was taken why not log base 2 or base 5 or any other number? Thank u

[@sunilsharanappa7721]

The log transform is a transformation technique to replace each variable x with a log(x), or to rescale your data in a logarithmic scale. A logarithmic scale is a way of represent extremely varied values (large range of positive values). It is based on orders of magnitude, rather than a standard linear scale, so the value represented by each equidistant mark on the scale is the value at the previous mark multiplied by a constant. The constant is the base, in this context. Considering x = 100: Base 10 - the base 10 logarithm of 100 is 2, because 10^2 = 100 Base 2 - the base 2 logarithm of 100 is 6.64, because 2^6.64 ~ 100 After the log10 transform, the increase of 1 unit represents x10 in the original scale. If you use log base 2, each unit will represent x2. Humans intuitively use the decimal system to count and do math, so it is easier to understand 1 unit as "10 times" than "twice" in this context. In practical terms, if you are not considering human interpretability, the any base will work just fine for your transformation. There is a nice post from Daniel with respect to your question which is worth reading, please take a look. www.linkedin.com/pulse/which-base-should-you-apply-your-log-transform-daniel-labbe/

[@shaghayeghraaaaad]

Thank you sooo much

[@karannchew2534]

Sir, A question re: log(income) use case at 5:35 please. While log scales down the super high income, it's also compreses other income amounts, most of them at same order of magnitude. Wouldn't that create a new problem: the log-ed incomes are now too close to each other? Wouldn't it impact the ML ability to predict results?

[@codebasics]

No it will not. Once model is built and when you give it an income value for a prediction, it will again take log of that input value. So overall it will work out and you won't face any problem

[@karannchew2534]

@@codebasics Thanks bro. Also THANK YOU for sharing your video and knowledge.

[@muhammadsajid4344]

Good job, sir

[@codebasics]

Most welcome

[@reserch1098]

Hi do you implement any hybrid machine learning models? collaborative filtering, rule based classification, and association rule coding....

[@nitinrahuja]

Thanks for this amazing explanation. Can we say that using Log function, you have normalized the revenue and income in the respective examples...?

[@codebasics]

Yes, you can say that.

[@dakshbhatnagar]

But isn't the min-max scaler given by SciKit Learn better for scaling the values where we are specifically facing an issue of 1 big number?

[@Satya-chowdhury]

Very good

[@rahulzinjurde2526]

Thanks sir

[@codebasics]

Welcome

[@nurulamin7982]

great job.

[@codebasics]

Thank you! Cheers!

[@anees001]

Hi It's a very quick informative video with practical example. 👍 I have a question, after we convert the revenues into log values, how to represent the y axis values in a report, so that everyone can clearlyinterpret the graph. Thank you

[@busrabintakalam7300]

Informative,🙂

[@codebasics]

Glad it was helpful!

[@Gurur_shots]

Sir how to calculate logy for salary column using python code????

[@ankushtaneja8118]

amazing

[@codebasics]

Thanks

[@boejiden7093]

I have one question. When you applied log to the salaries of the people in excel, would it make more sense to standardize it instead? Like when is it a good idea to standardize and when is it a good idea to use log?

[@codebasics]

Usual guideline is that when you have an outlier and skewed data (as indicated in the loan dataset in this video) it is better to use log transform. When you have two features on a different scale (i.e. bedrooms vs square ft for property price prediction) it is better to use regular scaling techniques

[@boejiden7093]

@@codebasics So in the example, you had credit score, age and income. Should I first do log transform to the skewed and outlier data like income and then standardize the columns? So should I standardize at all for that example? Thanks!

[@swarnachoppella388]

Thanks a lot : )

[@anandtrivedi6217]

cheera from Gujarat bhaai .

[@SahilAnsari-gl3xu]

Why you are using Log income instead of Normalised (0-1) income???? What is the difference in these two scaling on ML performance?

[@balajikartheek285]

hi sir, just a quick question, at the start of the video it says it will take us 2 years to get 5 dollars to 125 dollars at 5x return, as it will be 25 dollars after 1 year and then 125 dollars after 2...why does the answer become 3 years when performing the log function? Wouldn't that be like starting with 1 dollar, then having 5 after year 1 at 5x return, then 25 dollars after 2 years at 5x return and then 125 dollars after 3 years? Please help me!!

[@codebasics]

Balaji thanks for pointing that out , you are right. Actually I mean to start with 1$ and then after one year it becomes 5$, 2 years 25$ and then third year it will be 125$. I will add a clarification but you are absolutely right and thanks for pointing it out

[@prudhvi5211]

see when we have a data like single digit to 3digit value then we need to compare both then log scale was best base 1 to 10, another bar will be 10 to 100 was the same distance again 100 to 1000 in y-axis

[@adipurnomo5683]

So many machine learning algorithms using logarithm, right? E.g decision tree, logistic regression

[@learnerlearner4090]

Added below code segment import numpy as np df2 = pd.read_csv("income.csv") df2['log income']=np.log(df2['income']) I'm getting different values for 'log income' column. Pls. let me know if the log function used is not correct. Thanks

[@learnerlearner4090]

Got it. It's log10() that gives the values same as the values shown in the video. Thanks.

[@himanshusemwal1889]

Hey Sir! Plz help us on explaining the mathematics behind various gradient descent Algorithms and optimizations techniques. Probably the calculus part behind it .

[@codebasics]

yes i will be covering all of that in this series

[@puneetaggarwal7549]

Amazing

[@subhajitdas8871]

Please make a complete video course of Statistics and Mathematics.. required for Data Analyst or MIS job..

[@boojaado]

thank you

[@shamirrrr]

I have a question here. In loan approval problem you said if we apply log then values are more comparable and our model will not be baised. Then why we use min max scaler etc. Now i am confuse in between.

[@codebasics]

Very good question Shamir. Log transform is used when your data is skewed and you have outliers. In my example there was one person with abnormally high income and in this case it makes sense to use log transform. When you don't have outlier situation but say one feature is in range 1 to 5 (i.e. bedrooms for property price prediction) and another feature is in a range 2000 to 10000 (i.e. sqr ft area) then in this case it makes sense to use min max scaler. Here is the good article that gives details on what I just said above: medium.com/@sjacks/feature-transformation-21282d1a3215

[@shamirrrr]

@@codebasics thank you sir, That's the great explanation. ❤️❤️❤️

[@p.s_kev.in96]

I had the exact same doubt post watching the video now , but the explanation solved it 👍

[@TusharSuvarna]

thanks a lot!

[@codebasics]

Glad it was helpful!

[@gokkulvd9568]

Sir, your videos are great. we understand your effort but please take care of your health.

[@siddharthadebvines5296]

which one, data scientists should use ? spyder or pycharm ?

[@codebasics]

anything is ok. I use pycharm.

[@ran23111985]

Hi sir Is this applicable even for learning data analyst skills?

[@codebasics]

yes

[@shahpujanvinodbhai8926]

For people like me who don't know/remember more about log. Default base value of log is 10.

[@muzafarhussain6878]

absolutely gorgeous explanation❤. love you from pakistan

[@anoor7227]

Sir yh 10 video kafi for data sci science

[@mvcutube]

Thanks

[@codebasics]

Welcome