To determine an optimum path for the garbage collection truck drivers so that maximum resource utilization is achieved. Collected real-time vehicle data using OBD sensor and Raspberry Pi and sent it to Cloud for further analysis. Gradient boosting algorithm was used to predict the time taken between two points of travel. Genetic algorithm was later implemented to determine the final optimal route. Visualized the optimum route on Google Maps.

Determine from Twitter users the positive and negative sentiment of each party in state elections. Streamline data using Python and Twitter APIs filtered location-wise. Perform various data cleaning activities and sort data based on keywords. Ready to use word dictionary with sentiment values utilized for sentiment values. Final sentiments are party-wise evaluated. All the results are displayed with interactive word clouds which identify the most used words along with positive and negative words.

Region-wise unemployment prediction helps job agencies to focus on where to organize job events and hence an economic model was developed for unemployment prediction. Previous 5 years of data collected from the employment agency along with various economic indicators. Feature identification is performed to get the best features. Regression techniques like multinomial, SVM, and decision trees are implemented to generate the best model.

I collected sensor data from vehicles and detected anomalies in vehicles in real time. This data was stored in mongodb data base. Vehicle route optimization was also developed in addition for a logistics firm

Determine from twitter users the positive and negative sentiment of each party in state elections Streamline datausing pythonand tweeterAPI’sfiltered location wise Performvariousdata cleaningactivitiesandsortdatabasedon keywords Ready to use word dictionary with sentiment values utilized for sentiment values Final sentiments are party wise evaluated Allthe results are displayed with interactivewordclouds which identifies the maximum used words along with positive and negative words Python, Tweepy (Twitter APIs), Natural Language Processing Toolkit

As a part of Master Thesis convolution neural network algorithm is implemented on FPGA hardware for better hardware acceleration Image classification is performed on hardware with lower latency Convolution neural network is developed in Tensorflow (a computational tool in python) and then implemented on hardware using Vivado HLS(High level synthesis) using a C++ code An efficient image classification was developed with lesser process time on hardware