#10 ROC curve with AUC, Sensitivity & Specificity | Multinomial Logistic Regression in R

Рет қаралды 5,941

Күн бұрын

Week-10 R and data Files:
github.com/bkrai/Statistical-...
TIMESTAMPS
00:00 Logistic regression
02:00 Confusion matrix, Accuracy, Sensitivity, Specificity
05:15 Baseline rate
10:35 Effect of change in threshold or cutoff
15:40 ROC Curve
21:16 Area under curve (AUC)
23:00 Working with R
26:25 Confusion matrix from logistic regression model
29:14 ROC curve in R
34:30 ROC curve with AUC, best threshold, sensitivity & specificity
37:00 Best threshold & AUC
38:00 Two ROC curves on the same plot
43:10 Why logistic regression? Why linear regression doesn't work for factor type response?
01:00:39 CTG data & Multinomial logistic regression
01:03:02 Independent and dependent variables
01:03:39 Probability equations for multinomial logistic regression model
01:06:24 Model interpretation, coefficients, standard errors & 2-tailed z test
01:12:48 Writing the equation
01:15:06 Confusion matrix and related metric
01:15:41 multinomial logistic regression in R
01:34:30 Multi-class ROC Curves
R is a free software environment for statistical computing and graphics, and is widely used by both academia and industry. R software works on both Windows and Mac-OS. It was ranked no. 1 in a KDnuggets poll on top languages for analytics, data mining, and data science. RStudio is a user friendly environment for R that has become popular.
#MultinomialLogistic #MachineLearning

Пікірлер: 36

@asthamalhotra2345 3 жыл бұрын

Thanks to all your sessions I have successfully landed a job as a data scientist and moved ahead in my career. Thanks Dr. Rai. I aim to do a PhD someday in this field😊

@bkrai 3 жыл бұрын

That's excellent! Congratulations!

@abdulkhader101 3 жыл бұрын

You are a Legend Sir..You give all real time scenarios with Clear examples...Please keep posting KZfaq Vedios..

@bkrai 3 жыл бұрын

Thanks for comments!

@karthick2254 3 жыл бұрын

Nice! Thank you!

@bkrai 3 жыл бұрын

You are welcome!

@dimplekashyap1 3 жыл бұрын

Great video 👌🏼👌🏼

@bkrai 3 жыл бұрын

Thanks for the visit!

@Porusharda 3 жыл бұрын

Excellent video Sir 🙏

@bkrai 3 жыл бұрын

Thanks!

@smparulekar6103 2 жыл бұрын

Very Good ..Thanks

@bkrai 2 жыл бұрын

You are welcome!

@sallu.mandya1995 3 жыл бұрын

i got internship in ds , all thanks to you sir, god bless

@bkrai 3 жыл бұрын

That's great to know, Congratulations!

@adityachakraborty1814 Жыл бұрын

Very informative video. What is the null hypothesis of Mcnemar's test in this context?

@kar2194 2 жыл бұрын

Hi Dr Rai, at 31:53 regarding "Setting direection", how does the system know that 0 is controls group and 1 is cases group? Thank you

@bkrai 2 жыл бұрын

It's the default setting or this function.

@munthirmansour7554 Жыл бұрын

hello sir thank for your the nice videos, i am a cardiologist in the US, and does research in cardiac imaging. i have a question, when ever i do this code: plot.roc(r1) I get this: Error in roc.default(x, predictor, plot = TRUE, ...) : No valid data provided. not sure why i made sure i followed your steps and i tried to google this but no answer, could you tell me what i did wrong Thanks alot

@bkrai Жыл бұрын

If you are using the same data as in the video, then check your code again. If you are using a different data set, I may have to look at the code.

@soumyendupaul9556 3 жыл бұрын

Sir long waited for this video. Thanks a ton. Jus one query what's the significance of two tailed z test?

@bkrai 3 жыл бұрын

It's to see if the variable is statistically significant or not. If it is not, then it can be removed from the model.

@soumyendupaul9556 3 жыл бұрын

@@bkrai Sir what's the criteria to decide variables are significant under z-test?

@bkrai 3 жыл бұрын

If p-value is less than 0.05, it is statistically significant.

@asadalisarkar4253 3 жыл бұрын

Sir please make a tutorial on multiple polynomial regression in R

@bkrai 3 жыл бұрын

Thanks for the suggestion!

@sanjibphukan8921 2 жыл бұрын

Sir, I have a query, kindly suggest. I tried to build two models, MLR and SVM with training and testing data. The sensitivity for MLR-training is 62.9% and SVM training is 65.7%. Whereas, the sensitivity for MLR-testing is 70.0% and SVM-testing is 63.3%. Please suggest, why the sensitivity decreased in the case of MLR though it was high in the training data set as compared to SVM. And how we can to interpret this kind of situation.

@bkrai 2 жыл бұрын

It's difficult to say without looking at the data. But for choosing the better model, we should go with the results from test data.

@sanjibphukan8921 2 жыл бұрын

@@bkrai Thank you so much sir

@bkrai 2 жыл бұрын

You are welcome!

@datascience1274 2 жыл бұрын

Hello Professor, I'd be curious to see how to construct the plot for logistic regression with multiple variables. Can you maybe share the code with us, or suggest a video where you show this? Thanks in advance

@bkrai 2 жыл бұрын

This video includes Multi-class ROC Curves, if you are referring to that.

@datascience1274 2 жыл бұрын

@@bkrai No I was referring to the actual logistic regression model, with the x variables and the probabilities on the y axes. I have trouble to do that with multiple x variables. It’d be great to see that on the admit, gpa, rank dataset. Thanks a lot

@bkrai 2 жыл бұрын

I didn't understand your question. By 'multiple x variables' do you mean several independent variables.

@datascience1274 2 жыл бұрын

@@bkrai yes Professor. I'd like to plot each independent variable on the x axes and the respective predicted probabilities on the y axes. Im new to R, so Im finding this particular thing difficult. If you can share the code I'm sure it'd help me and many others. Thanks a lot

@bkrai 2 жыл бұрын

For making such plots in R, you can refer to this: kzfaq.info/get/bejne/eblxeJln0d3NqXU.html