Introduction

My name is Dinesh Kumar. I’m an inquisitive thinker and a Python Developer who has his commitments towards data science. I am dedicated and always eager to learn more and more about data science, machine learning and artificial intelligence related knowledge.

About You

I’m Dinesh Kumar, currently pursuing my btech from NIET greater noida in Computer Science Department. My Hobbies are to play Cricket and love to listen songs.

Education

• I have completed my High school Qualification from Ram-eesh International School in 2018.
• And currently I am pursuing my Btech From Noida Institute of Engineering and Technology in greater noida from Computer Science Branch.
• Currently I am in 3rd year and will passout by 2022.

  Projects

• I have prepared many models like Supervised models, Unsupervised model, Decision tree model.
• I have also done EDA on a mall details.
• In the month of october I also completed my Internship in Spark’s Foundation as Data Science and Business Analytics.

Why WOC?

After I completed my internship i think to search for more to learn extra in this field so I found about WOC(winter of code) and I read about this and feel that I should apply in this to learn about how to build various models and will get to interact to different people. So overall I am happy now after joining this and every day I learn new things.

Welcome to my Winter of Code Blog!

• Hello Everyone! My name is Dinesh Kumar. Currently i am pursuing my Btech from NIET in Computer Science department from Greater noida. This is my first week blog. I have been Selected for WOC 2020 at sub-org DSC-IEM and the project is to make model on TEXT SENTIMENT ANALYSIS url:-github.com/khanfarhan10/TextSentimentAnalysis.

Ideas

• Text Cleaning.
• Creating BOW.
• Applying n-grams.
• Apply various Machine Learning Algorithms(eg:-Naive Bayes,Logistic Regression).
• Enhance model performance and take it on upto mark.

What I implemented in this week :

Firstly i searched for dataset on which I will train my model. I found Amazon food reviews dataset then i import it and start doing EDA and done some Text Cleaning. I train it using BOW n-grams and TF-IDF and I used NAIVE BAYES and LOGISTIC REGRESSION Algorithms.

Upcoming Plans for Next Week?

• I will work more on to improve model performance and further firstly deploy it on local server using streamlit and then deploy it globally using Heroku.
• I will update UI for my model and make it more user friendly.
• I will make a Sidebar in which user can choose which algorithms he/she want to use for prediction.

Did you get stuck anywhere?

I stuck when I was training my model. Its performance was not coming upto mark so I searched and found about hyper parameter tuning so I used optimal alpha to find good hyper parameter and then trained it so now it comes nice one its accuracy score is 90.86% and F1 score is 0.81 and after that I also applied n-gram to enhance it performance.