[Day 23] Natural Language Processing 모델 만들어 봤다

 안녕하세요!
오늘은 Day 23입니다!


DeepLearning.AI의 NLP course에서 배운 것을 새로운 데이터에 적용해 봤다


첫 번째,

텍스트 감정 분석 하려고 Kaggle에서 재무표 텍스트와 감정 (positive, negative)으로 나눴던 데이터를 찾게 되었다

stopwords 없앴다


데이터셋을 나누고 토큰화 패딩도 했다
모델 구조는 다음과 같다

여러가지 모델 구조로 훈련했는데 대부분에 과대적합을 볼 수 있었다. 위에 모델의 성능은


과대적합이 텍스트 감정 분석 할 때 흔히 볼 수 있다고 했는데 극복하기 위해 더 노력해야 한다!


자...

두 번째,

텍스트 생성 모델을 만들어 봤다

이번엔 제가 만든 데이터로 모델을 만들어 봤다

1. 텍스트를 토큰화했다


그 다음에 각 문장을 기반으로 subphrase를 생성했다

예를 들어자면 I like dogs 문장은 > I - 1, like - 2, dogs - 3 

> [ [0,0,1],

     [0,1,2],

     [1,2,3] ]으로 변경했다


여기도 다양한 모델 계층 구조 해 봤는데 결국엔 아래와 같은 모델 구조로 했다


정확도


손실


그리고 마지막으로 ~ 텍스트 생성해 봤다

['Chelsea', 'Manchester United', 'Arsenal', 'Manchester City', 'Tottenham', 'Best team was', 'Against the wall', 'The FA cup was hard but', 'The season was filled with hardships and', 'In the end, the best team was', 'Some teams performed bad, so they were'] 

이 프롬프트로 모델이 그 다음 단어 20가지 생성하라고 결과는

1. Chelsea amidst a season of transition under the management of club legend frank lampard demonstrated an evolving form in the league
2. Manchester United with flashes of brilliance and frustrating dips in form in the lower ranks added an extra layer of intensity showcasing
3. Arsenal reached the final of the fa cup overcoming formidable opponents including manchester city and chelsea secure a chance to free
4. Manchester City led by the tactical genius of pep guardiola sought to defend their premier league crown with the same possession based
5. Tottenham amazon made a significant impact by broadcasting two rounds of fixtures in december featuring the highly anticipated merseyside derby occasionally
6. Best team was ensure an organized viewing experience games played on the same day were assigned separate time slots preventing overlap and ensuring
7. Against the wall season was characterized by moments of brilliance on the pitch as well as challenges that ultimately impacted city's quest for
8. The FA cup was hard but and son heung min and kane continuing to be standout contributors in attack maintained city's trademark attacking prowess led by
9. The season was filled with hardships and brighton exemplified city's dominance in disarray tournaments success in other competitions and wolverhampton while arteta southampton their modern history of
10. In the end, the best team was a mere sporting event securing eight consecutive wins 24 of which 7 were carried over from the previous season of
11. Some teams performed bad, so they were threats to opposition recognizing that klopp and ultimately brighter days lay ahead for the blues under their legendary manager that


내일은 시계열 모델 구성해 보도록 하겠습니다!


오늘은 여기까지입니다!

내일 뵐게요!

Popular posts from this blog

[Day 198] Transactions Data Streaming Pipeline Porject [v1 completed]

Image
 Hello :) Today is Day 198! A quick summary of today: data streaming pipeline project [v1 done] Here is a link to the project's repo. Well ... I did not know I can do it in a day (~14 hours) after yesterday's issues but here we are. Turns out in order to insert the full (~70 variables with nested/list structure), I need the proper pyspark schema. And yesterday I did not have that and that is why when I was reading data in the kafka producer I was getting NULL in the columns - my schema was wrong. Well today I not only fixed the schema for the 4 variables I had yesterday, but included *all* the variables that come from the Stripe API ~ 70 (for completeness).  When I run docker-compose, the data streams and is input into the postgres db (and is still running). Unfortunately, the free Stripe API for creating realistic transactions has a limit of 25, so every 3 seconds, 25 new transactions are sent to the db. It has been running half the day (since I got that set up) and as I am wr
Read more

[미리 공부] 기초 통계 복습 (Day 1는 1월2일)

Image
모험을 시작하기 전에 기초 지식을 복습하고자 했다. 오늘 SPSS 대안 프로그램을 찾아보려고 해서 JASP에 대해 알게 되었다. 유용한 프로그램인 것 같아서 선회귀와 기술통계를 내려고 했는데 재미있었다. 그런데 JASP에 대해 더 알기 전에 '기초 통계 지식을 좀 복습을 하고자 하면 좋을 것 같아'란 생각을 들었다. 다행히, Coursera에서 Stanford University의 Guenther Walther 교수님께서 진행된  Introduction to Statistics  무료 강좌가 있다.  좀 부족한 부분은 다양한 검정통계 하는 거고 (F test, t-test, chi-square 등) 이제 JASP 아니면 다른 통계 프로그램 사용하게 되어도 이런 부분을 좀 더 자세히 집중하여 공부하면 된다.  특히 homoscedasticity 및 heteroscedasticity 개념을 기억에 남았다. Homoscedasticity (선): Definition: In a homoscedastic dataset, the variance of the errors (residuals) is constant across all levels of the independent variable(s). In simpler terms, the spread of the residuals is the same throughout the range of predictor values. Heteroscedasticity (악): Definition: Heteroscedasticity occurs when the variance of the errors is not constant across all levels of the independent variable(s). In other words, the spread of residuals changes as the values of the independent variable(s) change. (Udemy 사진) 강의
Read more

[Day 61] Stanford CS224N (NLP with DL): Machine translation, seq2seq + a side CDCGAN mini project

Image
 Hello :) Today is Day 61! A quick summary of today: Covered Lecture 7 : machine translation, seq2seq, attention from Stanford CS224N Tried to make a conditional DCGAN to generate MNIST numbers ( colab ) ( kaggle ) I will first cover the GAN story (then will share my notes from the lecture) So... while watching and taking notes today, I started thinking, what if I can use my notes as data to a model and afterwards, when I want, I can give it raw string text and it will output text in the format of my notes (with my handwriting). Well I started looking around and actually the first model architecture that came to my mind was the GAN (specifically conditional GAN) - I remembered there was a GAN architecture that alongside the pictures, we can give it the labels, and then on-demand generate. In retrospect, there are of course others, but I decided to go with GAN.  For maybe 2 hours I busted my head trying to make a simple model with the EMNIST dataset (english characters), and I kept gett
Read more