How Can I Sign Nebraska Banking Form

How can i industry sign banking nebraska form simple

hi guys welcome to tech genius academy we are d industry-led training providers for those who wants to build their career in data science analysis iot device driver embedded linux and many more please do like share and subscribe to our channel [Music] hello everyone i am akshara shirvasa from techogenius academy let's continue the session of nlp basic concepts so now let's go to the next section of it which is a text cleaning part okay so before creating any features okay whether it's a normal machine learning model also right ah so before passing any feature into that so we need to understand what features are going to be important for uh deriving any insight out of it so some because this is a human generated text right so it might contain a lot of information which is not at all important uh for deriving any insight so it's uh useful that we have to adopt based on the data that you are handling we have to uh apply a couple of the text cleaning activities so these are the very basic and the generic text cleaning activities that i have listed out here uh before doing any feature engineering part on top of it so it's like the text can be written in any la uh any uh cases right it's kind of upper case lowercase and something like that so we don't want that uppercase word or the lowercase word should depict the different meaning although it is same right so uh in this way we can normalize the uh words so we just convert everything into a case then the punctuation punctuation sometimes don't contribute much so we can remove that as well then same thing with the numbers if you uh the numbers are not very important for your use case you can remove that then sometimes your text contains a lot of white spaces blank spaces so you can just remove it that as well then there are special characters as well sometimes you can remove that now some emojis hashtags are also there special words are also there if that is not contributing into your uh thing that you wanted to you know analyze so you can simply remove that um and then uh this replacing contractions with their expansions okay so this contractions mean sometimes we say something or write something like i haven't done this okay so that haven't so you can expand i don't want that haven't and have not should be interpreted differently so it's a good idea that what wherever the contractions these kind of words are there you just expand it into the original forms so that it should be comparable so but we need to understand like how what kind of a data that we are handling and based on that only we have to apply text cleaning it is not you know mandatory that you have to have apply all these exclaiming no because sometimes these special word emojis and punctuations right uh that is actually uh going to add importance into your feature generation because um there what happens uh some of the steps might not be suitable right for particular steps so uh like when you are getting the sentiments correct so these punctuation and word marks so these creates your emojis okay so you can't process that emoji because that emoji is going to tell you the uh emotion of a person it is going to identify what's a person's emotion or sentimented test so if you are going to remove it you are not going to understand what that text is all about right so based on our text and based on our requirement we have to apply the text clean so whatever i have discussed here i simply created a simple small small um functions out of it so these are the very basic one like a text lower here i've created and remove punctuations okay so i have imported these many libraries which are going to handle all the steps that i have covered over here lower punctuations remove the numbers and white spaces html sometimes html markups are also there right you can remove it and uh yeah contractions as i discussed so um this uh these you can you know call based on your requirement let me execute that so let me do one thing so this is my data right so let me take first text okay that is your first row so this is your row and let's uh apply all the um all the text cleaning one by one yeah yeah so this is my first row and i'm just trying to remove the punctuation so let's see how it looks like so this is your previous data like original data and this is after removing the punctuations so if you can see these things are gone right and then you have the punctuations and punctuations you have fullest job these hyphens these are actually gone so this way your this punctuation works now let's move on to the other one the lower case see the very first one also got converted into the smaller case here everywhere i got you know change alice here also it got converted into the lower case so this is how you can you know make it uh into the same case okay i think i have done remove punctuation two times let me remove that then remove numbers okay see here we do have this 39 something right so now we are not getting any 39 over here yeah maximum five days so now five is gone over here so that is what we need to understand whether that numbers are important for our analysis or not so if it is an important then you need not to you know uh execute this one um then we can remove the white spaces let me see whether we do have any white spaces are there okay so white spaces means if there are multiple spaces instead of only one space if you have multiple spaces in between right so this will basically remove that multiple spaces and just keep the one space within two words yeah and this html this i don't think so there are any html tags so maybe i can show you how it works okay html so here in this one you have to provide any text right hello this is the head sorry this is the head of your html block yeah so if you can see here we have this head only but since it is a html tag so it got removed yeah okay fine um here uh this last one uh replace contractions so i i don't think so again uh any um contracted word are there but i will show you how it works again with the our own example so i okay so that is a catch here so just use these brackets i have not visited that please yeah okay okay sorry i have given the wrong thing so see this haven't is just expanded into half not so so that we can you know normalize our text so somewhere if i have not as explicitly mentioned then if you are not applying this contraction so haven't and have not the machine interprets it differently not the same thing right so this is where you can use that then um let's go to the next topic uh which is uh the tokenization okay so this is the next again the uh very useful uh uh step of a kind of a useful pre-processing step on which other steps uh are dependent right so this tokenization is um as you understood the word token token means one a chart right so it is a process of dividing your uh large text into the smaller chunks so text can be divided into the sentences and they can further uh divide into the different two parts so we are going to um use the nltk uh package for this okay in order to understand a couple of these tips so nltk we are going to use which is a natural language toolkit nlp gaming's natural language toolkit and it is one of the best and well known and mostly use uh nlp libraries in our python ecosystem and it is useful for all sort of tasks like tokenization and stemming uh part of a speech which i am going to cover a little later and more uh any um like other activities for other activities as well yeah so let me uh start with this thing so i have imported this from this nltk package i have imported this word tokenizer it is going to uh tokenize this entire text so this is the given text and uh this is a method that you can imply uh word tokenize and provide this text so let me quickly execute it yeah so now you can see here how the entire text can be you know uh splitted into a single single word so first tutorial on these are the different words so this is how uh this word look nice to know is your words and then as i mentioned uh not only you can recognize your words but also the text also does into the sentences so if you have a large uh piece of text so the good idea is first to split the entire text into different different sentences and within that sentence then apply this word token so suppose this is a text that i am going to use okay this is the first tutorial on nlp the second text is next tutorial are coming soon okay and same thing same package i'm applying it but um this method is different scent tokenize is there just executed see this is the first tutorial on nlp this is your first token which is your sentence and this is your second sentence yeah in this way you can split your text data so there are other tokenization method as well so if you can see here within that analytical recognize only uh there are other also like uh using regular expressions you can you know split your entire sentence into the words okay using the punctuations or using the blank line okay so suppose i do have this text okay so it is going to uh you know perform all the three different uh types of tokenization okay so the first is it is going to do the regular expression i have given like you just split the text based on this regular expression okay need not to split on every spaces like here it was doing just uh split your text on base if it is following this pattern yeah that is your regular expression token then using the word punctuation and then if there is a blank line so that is why i have given this blank line let me execute it see so now here it is uh tokenizing it based on your thing whatever it has given over here yeah and then here it is doing it using the punctuation just see the difference here it is considering this as your one word or one token because we have given this thing over here in our pattern but since it's the punctuation one so what it is doing it is just considering everything as a different tokens so that is what we need to identify like which one is better for our purpose and then it is a blank like line tokenization right so wherever you have a blank line just split that thing so this is how it is happening so blank line is if you see here we do have over here so this is my blank line our next line so this is how it is splitting it up yeah so then we do have a another set of tokenizer using a tree bank so basically tree bank is a plastics corpus that annotates syntactic or semantic sentences structure so we have again imported the tokenizer and detoxer also so once you tokenize the sentence if you wanted to you know ah club all the tokens together into a proper statement that also you can do here so if you see here i have given one text over here and i've created a instance of your de tokenizer and tokenizer and then these are then i'm calling this tokenize on the sentence s and i'm just printing it out whatever what is the original sentence which is two right d tokenize is going to be uh done by this detoxinize method and then using a de-tokenizing using certain split thing okay let's do that yeah so your original sentence okay got to know something like this okay and and detox all the tokens get clubbed together and it forms your sentence yeah and um here in detox nice you can pass like how you wanted it to be this took nice thing uh okay so it is like uh create a different different sentences based on your space yeah so wherever it finds some space it's create a different sentences so this is uh the another step and then we can write our custom definition so it is like a definition and you are just calling it so now since uh till now here i was executing it on a single statement so now you wanted to execute on your given data set that we were working on that is your review data so that is why i have created one function simply and i am just putting an apply function which applies this custom tokens on each and every uh rules of the data set so this is the way it works i will show you how it looks so i created a different term column as well custom token so see till here we were having our original data and now we have created the custom tokens yeah so here it is getting stored in a list so this is how you can go so um let's continue the basic session starting with the stop words so now what is stop words so stop words are the words in the text which do not contribute much in categorizing to document from each other because they are very common occurring words so um i will show you what kind of stop words are available so different nlp based libraries provide you the stop word list that you can use directly use and you can customize it also so let me show you again same we are going to use the nltk and from here we are importing the stop word which is of english language so let's do that and let's see what you get in that list first of all i want to show you what is the type of that so just see what the type it is it is of list right um now just see stop word contains these words so as i rightly mentioned so we don't want uh such words into our vocabulary or into our analysis which do not contribute much into our into making any decision or into uh segregating two documents right so these are the words which are like common occurring words that doesn't con that doesn't differentiate two documents so it is better that you can remove these words from your text apart from you know removing uh the words from here you can also add couple of new words which is not available here which you want to remove it from your text so here what i have done it like some of the words which i do not want to be available into my text so i just wanted um these word to be just get appended into this stop word list so when i'm going to execute a removing of stop word method it is going to remove it from my text so simply you can just uh use the appender function of your list it is going to append it like new like i have added it if you can just see and search is new like you might be able to see somewhere yeah you can see new has been now added into your stop word list same way uh the same way you have added your new words like sometimes if you want to remove some of the word from this stop word list means you or you you do not want that these words should get removed from your text you want those words into your text so suppose uh in this issue of word right i have this because okay so i'm thinking that because it's an important word for or differentiation let's assume that so you can do the same thing like this yeah and let's see whether are you able to see the because word over here see it is not here yeah so in this way you can customize your stop word list whatever you want in your text to be there and what you do not right so now let's see how this is going to work so this is a word tokenizer method that i have taken and i'm just passing the sentence s so let me just see what is there in s yeah we do have this right so in this one okay this is a tokenized word so it is going to create a token and that then we are applying uh this is stopwatch removal using this list comprehension so what i'm doing for every word in your text token in your this okay if that word is there sorry if that word is not in your stop word list then consider it otherwise remove that so if you see off here we were having off and in so off and inward there in our stop word list so this function has removed off and n because they are not much contributing to our text yeah so till now we were talking about nltk so there are another popular libraries which is spacey which we will uh deep dive into our subsequent um session but i just wanted to show you uh that library also has their own stop word list right you can see here so these are the stop words so you can compare and contrast what all stop words over there and an ltk and whatever is there in your species so you have to load the model and then execute this default stop what's available so you just have a look on it okay so now um in order to apply this stopwatch removal into our data that we were handling for the reviews patient reviews i again created a function for it it is same as say same list comprehension i have taken it but put in uh kept it in a function if you see here my data is just what token is equal to my data this is a custom token that i b cause you know write a stop word works on the tokenized data so already we have this column custom token that we created uh during the tokenization step so just apply this stop word removal and let's see see uh we have is right it has now removed two that has also been removed so this is a wave uh by which you can remove the unnecessary word which is not very much contributing in the context yeah okay now let's move on to the next technique so the next technique is your stemming and limitization so what happens sometimes we do not require all the form of words like past chance present and the future we just want to know the base word of it so example we have like play playing and played right all three refers to the base word play so uh we are only going to be interested in that like in some search engine it doesn't require all forms so we need to chop the words uh into it original or base form so it just need base form in order to search anything it doesn't uh need all the forms to be present in your vocabulary yeah so limitation and stemming actually uh does this this job for us so if you can see uh we have this is a normalization techniques because we are normalizing the word into their base form and it is putting it into their canonical ones and as i mentioned it is going to be used in the search engines so now what is it in difference between the stemming and limitization so the main difference uh is that stemmer stemming it you know uh works basically on a single word and it doesn't uh consider the context uh of that word okay like in what context that word has been used in despite of that it just simply takes one word and then uh chop the ending of that word if if we have playing so it removes ing if we have does it remove es at the end so without understanding like a previous understanding like in what context we used it it uh doesn't uh consider that so and uh the other thing is uh since it is not uh you know uh doing that calculation how a word is used being used in the context so it is a little bit faster as compared to the uh uh limitization so again here also we need to understand like which one is best for our use case for our sort of data so these are the following uh like our different types of this demos that are available borders stemmers snowball um lancaster and regex yeah so these are the main features of each stemmers like which one is uh does what so here if you see it is a very aggressive one so it is then snowball is the improvement over the porter okay and it is computationally heavy actually so this one is faster as compared to this one and um lincaster is a very kind of or not very intuitive so it does an aggressive one simply remove the last word so you you you will get very shorter word which is a bit confusing you won't be able to understand what that word means sometimes so it reduces your work which doesn't uh good for our vocabulary generation yeah so uh the best for now we can consider it's like a snowball and also sometimes we use a portal so and rejects expression also so you can also provide your regular expression in order to stem your words what do you want uh to be considered while stemming uh the words to come up into your base form so if you see here i have given you an example between the limitation and stimming how it does have uh doing it so was studies and studying so see how stemming is simply a chopping of the last word s e s i n g but uh on contrast uh limitization is uh you know maintaining the context as well so this is uh let's see how it can be implemented in python so i am showing you all the examples for this chamber all the portal snowballs lincasters and rechecks yeah and this is the text one which we are going to apply the stemming and these are the different instances that we have created and then simply again it also works on the tokens right so again unless comprehend comprehension in order to uh apply the schemas different schemas on our tokens so let's execute it now here you can simply see the differences between the diff different stemmers so as i mentioned the casters how it does right uh simply uh just like got reduced to like i like it which doesn't make any sense right so in this way that's why it is kind of reducing your words then snowball and porter if you have if you see here yeah uh this is uh more or less giving you um the uh similar result but if you see here how it has you know uh maintaining the word uh meanings as well uh it is giving you has but it is like etching so then the next one is the rejects as i mentioned here if you see uh in rejects while creating the rejects i'm saying that whatever the word ending with ing just put it into the base form just remove ing there okay so uh in the text we have eating right so here simply it removes it simply keeps the heat so in this way you can also you know uh generate your own stemmer if uh it is very much particular to your uh data set then this is the limitation so if you see here in limitization so in limitization i have taken again from the nltk library word lit net uh limitizer it is also doing the same way using the list comprehension worked on uh these tokens and get the lemma off each um words so this is the output that you can see and just you can compare it with your stemmer how different it is from the different estimates that we have seen so here you can see apple we have got chopped off into a wpl so l e has been removed from here but here we do have a proper word apple because looking at the context of the uh sentences it is uh maintaining the proper words so this is how your the stemming and limitizing limitation differs while you know creating the base words of the different forms of word yes um now let's apply a limitation on our uh review data that we were having so same uh i'll use the apply function and just uh use the lemma using the list comprehension let me just quickly see how it looks like yeah limit is choking so here it is this is the thing that you can see here so um we have to cover the other topics as well so i will cover the rest of the topics into our next video till then stay tuned and please do like share and subscribe the video