Logistic Regression Using Excel

Predict who survives the Titanic disaster using Excel.

Logistic regression allows us to predict a categorical outcome using categorical and numeric data. For example, we might want to decide which college alumni will agree to make a donation based on their age, gender, graduation date, and prior history of donating. Or we might want to predict whether or not a loan will default based on credit score, purpose of the loan, geographic location, marital status, and income. Logistic regression will allow us to use the information we have to predict the likelihood of the event we're interested in. Linear Regression helps us answer the question, "What value should we expect?" while logistic regression tells us "How likely is it?"

Given a set of inputs, a logistic regression equation will return a value between 0 and 1, representing the probability that the event will occur. Based on that probability, we might then choose to either take or not take a particular action. For example, we might decide that if the likelihood that an alumni will donate is below 5%, then we're not going to ask them for a donation. Or if the probability of default on a loan is above 20%, then we might refuse to issue a loan or offer it at a higher interest rate.

How we choose the cutoff depends on a cost-benefit analysis. For example, even if there is only a 10% chance of an alumni donating, but the call only takes two minutes and the average donation is 100 dollars, it is probably worthwhile to call.

[nerygabrielnery]

Very useful video. Thank you very much. But there's a little detail people should be warned about: in order for the calculation to be performed correctly, every categorial variable needs to be transformed into a dummy following the Excel setup, which is a column for each dummy and all the others kept equal to zero.

[steinberg]

Hi! Ive downloaded and installed the real statistics tool pack for excel 2016 so my selection window, in video at 3, 32mins, looks different and im having problems understanding what option to choose. In the category of logistic regression i have 'binary and Probit logistic regression' and 'multinomial logistic regression' to choose from?

[pinkosheepo]

this is amazing. i know close to nothing and was able to do this and understand it

[Little_Alpine_Girl]

Hi, if i have 3categories in the output variable. What shall be the cut off? Shall i just go with 0.5?

It will be highly appreciated if someone could help me with the answer!

Thanks!

[absanamimi]

How to do flood mapping using logestic regression?

[garzv]

super useful, saved my thesis! thank you very much!!

[balaji]

Can you please post a follow up video that explains the rest of the results . I am using logistic regression to predict if a patient would be readmitted to a hospital given some variables. How would I apply the results to a
 given patient. Thanks

[lennon]

The Most important thing is Before open xlam, you should make sure that you checked the "Solver Function"
and than close the excel, finally open it from dialog . I also barely fix the Problem with "Password thing"

In site you can check this words "Most important, remember that Solver must be installed.
To check this, press Alt-TI and make sure that Solver (as well as RealStats) is on the list of addins with a check mark next to it. "

[alex_n_m]

Very helpful add-in and explanation!
Thank you!

[EDUComputing]

Thank you for this tutorial. Excellent explanation without using any technical words

[joshman]

Thanks, good video and the explanations and examples for each thing are really helpful. Is there a way to make the accuracy increase?

[veronicad.]

can you do the same thing for like a correlation? sorry i am really helpless idk what to do with my thesis!! thanks!

[nukestrom]

So how do you transform from test to train?

[dolosofamili]

Hello, I have a question. On the bottom part where we see the coefficients and Wald value and the P value. Do we remove the variable that has more than .05 p value? Same like the linear regression?

[attitude]

Any way to get Odds Ratios from the excel linear regression program?

[sugarrush]

Hi Lucas,

I've used your excellent add-on to predict market movements based upon data I've recorded. As expected, my parameters offered little predictive qualities, oh well.

I then decided to try and prove to myself the power of this method of regression. I created a two column data set (200 rows). The first column contained a random number between 0 and 100, my predicted value was set as 1 if the random number was greater than 75, otherwise 0. I expected the tool to be able predict with a high probability that numbers over 75 are predictive of the value being 1.

When I ran the Binary Logistic Regression, I got all values in the output tables set to #VALUE! any idea where I have gone wrong ?

Thanks

[MAILAQASW]

Thank you for the detailed explanation. very helpful.

But I alsos have a question?

I was able to replicate the steps and get similar results. How do you explain row number 549 that has 2 Success counts, 1 Failure count. totalling upto 3.

Is it possible?

PLese help.

Thank
Dev

[santiagoe]

Is there a way to weight the dependent variable using this package? Thank you!

[anantmehta]

Your video and website is brilliant. Thanks.

[raywinovah]

I have downloaded the real statistics tool pack but I cant see logistic regression. Please confirm if its the same as binary regression under regression.