-
Hypothesis Testing on Patient Data: Analyzing Patient-level data to deliver efficient diagnostic and treatment processes.
Methods Used: Shapiro Test, Mann-Whitney Test, Levene's Test, Chi-square Test, ANOVA Test
-
Predicting the Chance for Graduate Admission: Predicting the chance for graduate admission, along with understanding what factors are important for it, and how these factors are interrelated among themselves.
Model Used: Linear Regression
-
Estimating the Delivery Time of an Order: Estimating the delivery time of an order on the basis of what the customers are ordering, from where and also the delivery partners.
Models Used: Random Forest, Neural Network
-
Precting the Creditworthiness for Loan: Predicting the Creditness of MSMEs, as well as individuals, for loans to bridge the credit gap in the rural market.
Model Used: Logistic Regression
-
Predicting Driver Attrition: Predicting whether a driver will leave or not for ride-sharing company.
Methods Used: Ensemble Algorithms - Bagging and Boosting models, SMOTE
-
Anamoly Detection of Hydrogen Fuel Cell: Analyzing the aging process, charge, and discharge cycle of Hydrogen Fuel Cells and isolating the anamolies.
Models Used: Isolation Forest, Local Outlier Factor, and Elliptical Envelope.
-
Forecasting the Number of Page Views: Forecasting the Number of Page Views to optimize Ads placement.
Methods Used: Stationarity Test (DF & ADF), ACF & PACF, ARIMA, SARIMAX, Prophet.
-
Jigsaw Unintended Bias in Toxicity classification: A deep learning model that can classify the toxicity of a comment posted on an online platform. This is useful in identifying and tackling the menace of online bullying, and hate.
-
Visual Question Answering: A deep learning model that generate answer to an open-ended question about an image.
-
Personalized Movie Recommendation: Creating a Recommender System to show personalized movie recommendations.
Methods Used: Pearson Correlation, KNN, Cosine Similarity, Matrix Factorization