Hello Diogo. Nice explanation! 👏👏 I have some questions. 1) How to obtain explanation for a particular test value. For example, Input Instance of a test data Jay India 80 years old Now let's we wanna predict the COVID-19 risk. Output: high risk (let's It's a clasa) Explanation of being risky : From the model. Any information in your reply will be highly appreciating. Thank you
Hi @Diogo, great video. Covered all the points I had in my mind. Yet, I have a question, Is it possible to apply LIME to a Cosine Similarity model? If yes then how?
Hi Diogo, Yes you are correct. Cosine Similarity is pretty much already interpretable. What am I looking for is to maybe give the results a better visualization using LIME. Thank you for your reply and effort. :)
Hi! Thanks for such a great video. But is there an opportunity to use Lime without training data? I only have saved model and test values; how can I interpret samples in such a case?
Hello sir, I have applied this Lime code on my liver prediction dataset it executed. but , when I give location as "x_test.iloc[0]" It does not show the result of [0]th location record . please help me to solve this. thank you.
Hi Abubakr, thanks for the question. Unfortunately, no. LIME works best with classification problems and does not have a lot for Deep Learning or Time series. You can find more here: github.com/marcotcr/lime
@@DataHeroes Thank you so much for giving me this useful information, please, I need you to help me to solve my problem if you, I'm building the spam email detector classification with Randomforest classifier and I'm trying to use LIME to explain my prediction output but I have a mistake. please email me if you can by means of this email: abubakrhassanahmed@gmail.com Regards
Why the negative and positive contribution of these 5 features do not sum up to 100%? 0.07+0.06+0.05+0.05+0.03??? Even if i retain decimals, obviously this does not sum up to 100%
Diogo, how do I get the local fidelity of the model? Thanks
Another question: Are you Brazillian?
Hello Diogo. Nice explanation! 👏👏 I have some questions.
1) How to obtain explanation for a particular test value.
For example,
Input
Instance of a test data
Jay India 80 years old
Now let's we wanna predict the COVID-19 risk.
Output:
high risk (let's It's a clasa)
Explanation of being risky : From the model.
Any information in your reply will be highly appreciating. Thank you
@@DataHeroes Thank you!
Hi @Diogo, great video. Covered all the points I had in my mind. Yet, I have a question, Is it possible to apply LIME to a Cosine Similarity model? If yes then how?
Hi Diogo, Yes you are correct. Cosine Similarity is pretty much already interpretable. What am I looking for is to maybe give the results a better visualization using LIME. Thank you for your reply and effort. :)
Hi! Thanks for such a great video. But is there an opportunity to use Lime without training data? I only have saved model and test values; how can I interpret samples in such a case?
You don't need to use training and test data, you can just use the combined data, if that is what you are asking :)
Hello sir, I have applied this Lime code on my liver prediction dataset it executed. but , when I give location as "x_test.iloc[0]" It does not show the result of [0]th location record . please help me to solve this. thank you.
Hi
Thank you so much for this video .
i would like to ask you, Can we explain a LSTM model with LIME?
Hi Abubakr, thanks for the question. Unfortunately, no. LIME works best with classification problems and does not have a lot for Deep Learning or Time series. You can find more here: github.com/marcotcr/lime
@@DataHeroes Thank you so much for giving me this useful information, please, I need you to help me to solve my problem if you,
I'm building the spam email detector classification with Randomforest classifier and I'm trying to use LIME to explain my prediction output but I have a mistake.
please email me if you can by means of this email: abubakrhassanahmed@gmail.com
Regards
Hii!!.... its great tutorial about lime xai. simple and easy to understand, Thanks for such a
great video
Thanks!!
Is there a link to access the code in this video?
Not this one, but you can follow the tutorial :)
Can you explain a prophet model with LIME?
@@DataHeroes thanks for your answer, that helped me a lot =)
Why the negative and positive contribution of these 5 features do not sum up to 100%? 0.07+0.06+0.05+0.05+0.03??? Even if i retain decimals, obviously this does not sum up to 100%