Featured
AWS Cloud Practitioner Certified
AWS Cloud Practitioner Certified
AWS Cloud Practitioner Certified
Cloud certified Jul 2021
Oracle SQL Associate
AWS Cloud Practitioner Certified
AWS Cloud Practitioner Certified
SQL certified Jan 2021
Beta Gamma Sigma
AWS Cloud Practitioner Certified
Beta Gamma Sigma
Honorary lifetime member
Academic Projects
Amazon Text Classification
- Cleaned data by removing unwanted columns(review id), all the special characters(!@), single characters from the start, substituting multiple spaces with single space, numbers(0-9) an stop words.
- Developed visually interactive exploratory analysis using Plotly Library
- Performed NLP task like Stemming, and Lemmatization
- Performed feature Engineering by converting text into count vectors and TF-IDF vectors
- Applied Naive Bayes, logistic regression, SVM, decision trees and random forest models on the data set to determine whether the review is helpful or not
- Selected the best model, Logistic Regression, based on its AUC-ROC Curve(accuracy not chosen as data was imbalanced)
Dow Jones Index Forecasting
- Gathered monthly data of the adjusted Closing price of Dow Jones Index from the year 2000 to present from Yahoo finance.
- Applied basic forecasting techniques (mean, naive, seasonal naive and drift)to advanced(Holt, Autoregressive model, and Arima).
- Based on the overall performance of models(taking into consideration 7 different types of error), the ARIMA(2,1,1) is considered the best model.
- Created a Shiny App to display the results in a visually interactive way.
Classification of Adults with Autism Spectrum Disorder
- Applied Logistic regression, decision trees and random forest models on the dataset to determine whether the person has ASD or not.
- Selected Logistic regressions with recall 90.93%; selection was not based on accuracy due to imbalanced data set and high cost of false negative
- Used K fold Cross validation to attain reliable estimates for training and testing
Follow me on!
