Echhit J. Email & Phone Number

Washington, District Of Columbia, United States

Washington, District Of Columbia, United States

Greater Pittsburgh Area

• Responsible for binary classification of Sleep/Awake using 9 different smartphone sensor time series data and Fitbit data gathered using AWARE Framework for 120 users.• Engineered features from raw sensor data into various descriptive statistics such as mean, standard deviation, count, percentiles, and so on• Aggregated them in 3 different time windows (10/30/60 minutes) for Sleep prediction and data analysis.• Preprocessed data by resampling and joining Fitbit sleep data with engineered features, dropping features/tuples with high nullity, Imputing/interpolating features to eliminate nullity, and visualizing tables for data overview.• Implemented a time-table like data imputation method that assigns and updates average values for various sensors throughout the day in the respective time-windows that captures cyclical user behavior.• Created a pipeline for tuning various classifier model hyper parameters which consisted of oversampling(SMOTE) on training data, scaling features, and GridSearch with K-Fold Cross Validation on Training set.• Accurately predicted Sleep/Awake targets to get a mean F1 score of .83. Inspected overfitting with Training and Test set AUC scores.• Adjusted predictions by using the threshold value that maximizes the difference between Sensitivity and False Positive Rate to better classify sleep classes.• Created Sleep Profiles that accurately outputs Bedtime, Waketime, and Sleep duration for users.• Currently Analyzing sleep disturbances among the users and adding it to the sleep profiles for a comprehensive analysis of day to day sleep patterns.

Greater Pittsburgh Area

Grading Object Oriented Java and Python for Data Analytics

Pittsburgh, Pennsylvania, United States

• Developed an Android application that uses a Pi network to help during a Tsunami crisis for Padang, Indonesia.• Analyzed Social network data to efficiently communicate and relay message within communities.• Performed Hypothesis tests to check if the ad-hoc system would perform better against without the system using T-Tests.• Analyzed demographics data to find best community leader candidates that would help and communicate effectively with the community• Cohered various features such as GIS, instant messaging, and profiles for efficient communication.• Assisted in implementing real-time data flow in the application before, during, and after a tsunami for safe evacuation.• Devised a socio-technical hierarchical tree to connect two-way communication modules between leaders and followers.

Master Of Information Science, Information Science/Studies

Pursuing Information Sciences Graduate Degree with a concentration in Big Data

Bachelor'S Degree, Computer Information Systems/Business Administration, 3.69

Activities and Societies: National Society for Leadership and Success