• Built credit card anomaly detection system with predictive models using Logistic Regression (93.65%), KNN (99.98%), SVM (99.91%), SGD (94.26%).
• Handled imbalanced dataset using Random Undersampling, Random Oversampling, Random Oversampling using SMOTE analysis.
• Clustered Data into fraudulent and non-fraudulent transactions using dimensionality reduction technique with T-SNE.
• Used Ensemble models like Voting Ensemble (95.36%), bagging with Random Forest (100%), boosting with XGBoost (98.34%) and AdaBoost (96.89%).

• Clustered documents using self-implementation of Kmeans to find classes of documents and built a document classifier using self-implementation ofK nearest Neighbours and Fuzzy KNN. Used pre-processing techniques like tokenization, stemming, lemmatization, sliding window, NER, Ngrams anddimensionality reduction techniques like PCA and SVD.

● Performed aspect level sentiment analysis on Amazon ecommerce reviews using a lexicon-based NLP that helped businesses identify the features ofvarious products to concentrate upon for improving sales.
● Built rule-based model with negation handler and Tkinter based GUI that dynamically analyses the customer review input using context dependencies with the help of Stanford NLP parser for opinion aspect mapping. Performed Topic modelling using the clustering algorithm Latent Dirichlet Allocation.
● Developed web portal with visualizations like radial tree, scatter plot etc. for a hierarchy of managers and product recommendations for customers.

● Performed web scraping from any news website link in the configurable based on the category of news, dates and data processing.
● Included text summarization and sentiment analysis of each news article along with visualizations like boxplot, violin plot, bar plots, Word clouds,Time Series Analysis to understand the distribution of news data better to derive KPIs (Location, Timezone, Categories, Tags) for business

• Developed banking application using concepts like concurrency, fragmentation, transparency, deadlock handling, 2 phase lock using stored procedures, so that users could perform transactions in based on location.