Use case. I may have an actual use case for this. In my courses I give so-called fill-in-the-blank(s) questions. Students frequently misspell words in the most inventive ways possible (not by design of course) and I am pretty flexible in terms of giving full credit for "near misses". however, sometimes I wonder "How close is that answer actually?" This lesson gives me some ideas of how that may be accomplished by calculating the LD distance. The course management program I use (Blackboard) is not nearly good enough to do this however by itself. I would have generate versions I accept based on LD and feed those versions to Blackboard myself. I thought I would share these thoughts of mine. Thanks for the wonderful videos.
Hi, excellent video!!!! What is the default method for measuring distance for function "stringdist" here? Since you didn't set the method, I was curious.
Hey Jenny, thank you very much for the kind feedback, glad you like the video! The default method of stringdist is oas. You can find more info on this here: www.rdocumentation.org/packages/stringdist/versions/0.9.8/topics/stringdist Regards, Joachim
Excellent video. Very interesting stuff! I do have a request or suggestion. Kerby could you do a video or a series of video on NLP (Natural Language Processing)? It seems to be a field that is gaining steam. My son is a layer and a data scientist who studies NLP for legal docs and I would love to know what he does for a living.
Awesome tutorial. Levenstein distance still doesn't beat speed of fuzzyLookup in excel which is a shame. Neither does fuzzy join package. Frustrating bottleneck for automation but the performance is unquestionable. Tokenized jaccard in fuzzyLookup in excel still the king.
How are you viewing the actual values ([1] "Bill Clintion" "Barack Obama") rather than just the numbers ([1] 5 3) in this? I see you switch back and forth a bunch of times but I'm not sure how you're doing that.
Hello Tilda, You can use the value=TRUE argument in the use of agrep() function. It would give you the exact values or use the amatch() in square brackets to identify the index positions in the pres_df data frame. The script is given below the video. You should click on show more to see it. Regards, Cansu
Thank you! I have never done this myself, but this Stack Overflow thread seems to discuss your question: stackoverflow.com/questions/58718287/fuzzyjoin-with-dates-in-r
Hey Andrea, sorry for the delayed response, I was on vacation and couldn't reply earlier. Could you please explain your comment in some more detail? I'm afraid I don't get it :) Regards, Joachim
Thank you su much ! What you explain is exactly what I was looking for to deal with my data !
This is great to hear Loancet!
Use case.
I may have an actual use case for this. In my courses I give so-called fill-in-the-blank(s) questions. Students frequently misspell words in the most inventive ways possible (not by design of course) and I am pretty flexible in terms of giving full credit for "near misses". however, sometimes I wonder "How close is that answer actually?" This lesson gives me some ideas of how that may be accomplished by calculating the LD distance. The course management program I use (Blackboard) is not nearly good enough to do this however by itself. I would have generate versions I accept based on LD and feed those versions to Blackboard myself. I thought I would share these thoughts of mine.
Thanks for the wonderful videos.
Thank you very much for sharing this use case Haraldur! Indeed, this should be a good example where fuzzy matching is useful.
Hi, excellent video!!!! What is the default method for measuring distance for function "stringdist" here? Since you didn't set the method, I was curious.
Hey Jenny, thank you very much for the kind feedback, glad you like the video! The default method of stringdist is oas. You can find more info on this here: www.rdocumentation.org/packages/stringdist/versions/0.9.8/topics/stringdist Regards, Joachim
Excellent video. Very interesting stuff!
I do have a request or suggestion. Kerby could you do a video or a series of video on NLP (Natural Language Processing)? It seems to be a field that is gaining steam. My son is a layer and a data scientist who studies NLP for legal docs and I would love to know what he does for a living.
Thanks for the kind words and the great suggestion Haraldur! I'll forward it to Kirby. Regards, Joachim
Awesome tutorial. Levenstein distance still doesn't beat speed of fuzzyLookup in excel which is a shame. Neither does fuzzy join package. Frustrating bottleneck for automation but the performance is unquestionable. Tokenized jaccard in fuzzyLookup in excel still the king.
Hey Robert, thanks a lot for the kind words and the additional info!
@@StatisticsGlobe love your vids bud and your no bullshit approach. keep it up!
Thanks mate! :)
How are you viewing the actual values ([1] "Bill Clintion" "Barack Obama") rather than just the numbers ([1] 5 3) in this? I see you switch back and forth a bunch of times but I'm not sure how you're doing that.
Hello Tilda,
You can use the value=TRUE argument in the use of agrep() function. It would give you the exact values or use the amatch() in square brackets to identify the index positions in the pres_df data frame. The script is given below the video. You should click on show more to see it.
Regards,
Cansu
amazing!
Hey Jelly, thanks for the positive feedback! Glad you like the video!
this is great, would fuzzy_join work with dates?
Thank you! I have never done this myself, but this Stack Overflow thread seems to discuss your question: stackoverflow.com/questions/58718287/fuzzyjoin-with-dates-in-r
So _amatch_ is the most general function here for fuzzy matching
Hey Andrea, sorry for the delayed response, I was on vacation and couldn't reply earlier. Could you please explain your comment in some more detail? I'm afraid I don't get it :) Regards, Joachim
Hello! If I want to do an exact match and a fuzzy match at the same time how can I do it? 🥺
Hey, I'm not sure if I understand your question. How would this work theoretically?
nice
Hi Olphy, thanks for the comment! Glad you like it!
What do you suggest for a large data? (About 600,000)
Hey Paul, have you tried the code of this video? Did you get any error messages?