Diabetes Prediction using Machine Learning from Kaggle

Neeku dandalu dora! You're totally awesome man!

[@ShortVine]

i love this vid, i have watched 2 this kind of vid on ur channel, one humble request, can you please make more these kind of vid, because this is really helpful & important for beginners like me. Much love

[@eeshsingh3336]

Hi, thanks for the video. One question, is it possible that replacing all the missing values with mean is affecting the accuracy? As Insulin has a lot of 0 values and it is the main feature which can affect our final response in diabetes prediction. Is there a better way to impute the value so that it is more uniformly distributed

[@yusufvan]

Ya good question, i have the same question, help me with the solution

[@ijeffking]

Nice one. Your pointers are very useful. Thank you

[@ronakjain7192]

Why didnt you explain Hyper parameter tuning code? I guess most of us didnt understood in params section.

[@jaysoni7812]

parameters are basically mathematical stuff so it was already explained in particular algorithm's video check it

[@skh7056]

@@jaysoni7812 hey, what is the point of calculation of correlation of features..? Here, we even didn't do dimensions reduction?

[@shervintheprodigy6402]

I have a doubt, there is the independent variable named skin in his video but the actual dataset does not contain that independent variable, why is that? Am I looking at the wrong dataset?

[@Elif-tt8ez]

Why didn't you look at the missigness correlations when filling in the missing values?

[@syedmunazzirahmed3191]

Do you think accuracy of 77% is a good score ? Or there is potential to get better results?

[@tilakbhujade4438]

I worked on the same dataset and I have an accuracy of 81%

[@vikasvk9174]

your content is osm .....but pls explain it more clear at some point it is difficult to understand .....Thank you :)

[@ajaykushwaha4233]

Requesting to create few more projects in Healthcare domain.

[@Piyush-pj2od]

Hi there, It seems that value 0 in num_preg has also been replaced by mean. Value of num_preg can be 0 in the data. Could you please clarify?

[@theultimatetruth7087]

do you got the ans ? let me know if you

[@shivaniyadav7021]

how to impute missing values by using multiple imputation through chained equation.because of missing values in insulin and in some attributes it affects the accuracy of the model

[@pravinwakle131]

Why are you using mean imputation strategy? Any idea to check pattern in missing data

[@dhiranshsaxena7409]

I would suggest you to try it with KNN once. I tried KNN with the same dataset and achieved an accuracy of 81+

[@PatnalaYuvaMahalakshmi]

Can u share me the code with knn please

[@skh7056]

what is the point of calculation of correlation of features..? Here, we even didn't do dimensions reduction?

[@kissuist]

Null values are in the set

[@shahariarsarkar3433]

Brother how to write a research paper? Please help me. If I do any project that is shown in your videos then I write a article, will it be accepted as a conference paper?

[@nani4027]

Bro..here we are getting accuracy as a output..but..where is diabetes prediction

[@isko_vlog6836]

good job

[@soumyasrm]

Pls share a project detail on insurance fraud claim analysis ... end to end

[@krishnaik06]

Sure...probably the next video I will upload will be in that topic.

[@zaveriamutwalli6399]

Why you choosed random forest algorithm for this?. Actually i am new to this so i want to know reason of using this algorithm as compared to other algorithm

[@adiflorense1477]

4:17 I think if the label or class is in the form of a number it is called regression. is that so sir?

[@shervintheprodigy6402]

no its not true

[@adiflorense1477]

@@shervintheprodigy6402 why not sir?

[@shankarchintapalli3374]

How uci diabetic dataset is converted into attribute wise dataset

[@shreyasb.s3819]

Thanks.nice one

[@oratoradda8235]

I have my own data set with approx 15 parameters , and want to write a paper but not aware with data science , please suggest ....

[@busracelik3059]

which algorithm did we use in this video ? decision tree ? SVM ?etc.

[@GC-dh1zp]

Random forest and xgboost

[@ganeshbasalel5595]

could you please teach about Hadoop map reduce k-means clustering (H-KC)

[@Mayurtechmaster]

Can you make one video related to handle a high cardinality in a feature

[@ayoushdas8715]

how to differentiate between diabetes type from pima india dataset?

[@starab6901]

Can it be done by svm

[@PankajVerma-zj4vr]

You have a great communication skills Can you give any tips how to develop.. Please🙏🙏🙏

[@TechBrain811]

How will you handle the missing data if it is present?

[@vaibhavgaikwad1938]

Thanks Krish, I am trying to do my dissertation in healthcare analytics, wanted to know if you have done anything in quality care mining?

[@vijayashrivastava3383]

Hi I am not able to import Imputer from sklearn preprocessing I am getting the below error: ImportError: cannot import name 'Imputer' from 'sklearn.preprocessing' (C:\Users\Vijaya\anaconda\lib\site-packages\sklearn\preprocessing\__init__.py) Please do help.

[@gauravbogar5044]

use this from sklearn.impute import SimpleImputer imputer = SimpleImputer(missing_values=np.nan, strategy='mean')

[@dc09kaa50]

How to find diabetes prediction function?

[@dilipyadav1264]

Sir can our give any reason why we have used xgboost algo for improving accuracy why not another algo

[@LikhithaH]

Hi sir I am Likhitha currently 11 years I am getting a error that says "No module named 'xgboost'" please do help me out to solve this error thank you Likhitha

[@sunilkumarkancharla3517]

install package xgboost pip install xgboost

[@kumarajay7th]

Pls mention the significance of correlation of features found using heatmap matrix. Also how did u reduce the feature from 10 to 8 and why?

[@dhirendrasingh6071]

The rest two features were categorical whose correlation can't be found with numerical values

[@reshmajames6]

Can you say which algorithm u used !pls

[@sunnyarora4916]

after the imputation step, how do I see the mean value which is replacing all 0s in x_train ,how to see that?

[@kothapallysharathkumar9743]

Hai sir please make video on lda topic modelling

[@sunnyarora4916]

why does it give a different accuracy value when the data set is same

[@jaysoni7812]

bcz when we test the accuracy using test data set, some of types of test data dosen't include in train dataset so model doesn't train or aware of that data so that's why it give different accuracy, for fix it we have another method called KFold Cross validation check that video

[@upendra8050]

Thanks for the video. But your precision and recall are so high which is not good for this kind of analyses (because we don't want to wrongly predict a disease to be diabetic and also we don't want to miss any patient with diabetes. The precision is somewhat acceptable not low recall). Do you an intuition of how you can improve those and overall accuracy? Maybe some feature selection and feature engineering? I want to know more thoughts on it. Maybe a follow-up video on this?

[@noecareme8242]

I think more ellaborated feature engineering would do. I find featuretool package, for automated features engineering to be the best. I have had amazing results so far at deployment stage.

[@upendra8050]

@@noecareme8242 Great. Will try it out

[@skh7056]

what is the point of calculation of correlation of features..? Here, we even didn't do dimensions reduction?

[@chetanalate4038]

Ye project banake milega kya.....

[@noecareme8242]

Hi Krish. I enjoy your videos. I just wonder why you always use correlation to select your features. Yet, correlation only picks up the linear dependency. What if you use Wrapper methods or embedded methods? Thanks.

[@susanth1888]

Can i use it for my project? (Probability Paper)

[@sayanpaul2083]

if i want to take the input manualy to predict inplace of X_test how can i do that?

[@dhirendrasingh6071]

Just pass an array of your preferred values in place of x-test

[@riturajpurkayastha1774]

@Krish Naik Which ML algorithms you use in this project

[@sanketthore4384]

random forest classifier

[@jaysoni7812]

XGBoost

[@niftwithharsh4517]

What is the purpose of this?......is this regression you are doing on python??

[@niharikacram1434]

Sir why did u took only pregnancies patient only u could have taken others also

[@louerleseigneur4532]

Thanks Buddy

[@chetanalate4038]

Prediction of onset of diabetes using machine learning algorithms

[@sevanthiselvam1349]

Hii how to predict

[@nirajchaudhari5974]

I used logistic regression for this I got an accuracy:0.76663 and it does not take time for fitting

[@dhirendrasingh6071]

Lmao

[@riteshjain5059]

Is their anyone who import SimpleImputer instead of Imputer?

[@aqsaqamar1634]

Can you solve my one problem

[@kissuist]

Provide the code

I am getting 70.1% accuracy by catboostclassifier