Design and Development of V2V and V2I Communication Technology

In India, annually nearly 5,00,000 road-accidents occurs accounting to 1,50,000 precious lives. Hence, my team's aim was to reduce nearly 75% of such road-accidents occuring to multiple reason, which majorly caused by Driver's Fault, Road-Ragging, Mishap occuring at Blind Spots majorly and upon complete development would eradicate the oportunities for theives to steal our vehicle.

Key Technologies Used:

1. Arduino Uno
2. RFID
3. Zig-bee Communication Technology
4. Further development was proposed with the use of Data Mining, Implementing regression techniques.

E-Differential

It is an extremely futuristic project wherein, the entire mechanical system of power distribution to the rear wheels is promised to be replaced with the Electrical Drives which would results into reducing the great amount of power-loss as occuring in mechanically driven differentals

1. Arduino Uno
2. Speed Sensors
3. Speed Control Drivers
4. Future Development was planne with the help of Augmentation Reality of the Road-Condition's

Google Stock Price Prediction

It is one of the best regression model that I have implemented and practiced for practicing the application of keras, time-series plotting.

1. Model Used: Recurrent Neural Network
2. Dataset : Google Stock Price Data for 5 years

Hotel Customer's Review Classification

It is one of the most challenging model I have implemented for classifying the reviews given by the customers as good or bad(Feedback for Improvement), wherein I learnt the concept of Sparse matrix processing, keras, Re Library

1. Model Used: Natural Language Processing(Bag of Words Model)
2. Datasets : Reviews received from 10000 Customers.

Image Classification - Dog or Cat

The model was completely learning intensive and was my first , wherein I gt an awesome exposure to the use of Tensors, Keras library, and the most importantly how the convolution works.

1. Model Used: Convolutional Neural Networks(3 Convolution Outputs)
2. Dataset : Consist of 10000 Images of Cats and Dogs

Kaggle's Titanic Survival Prediction

Based on set of the inputs for the passengers voyaging in the Titanic in 1912, prediction for their chance of survival was expected. Herein, I learned practically by applying my knowledge of regression models.

1. Model used:
Naive Bayes, Random Forest, XGBoost, AdaBoost, Support Vector, and an ensemble model of them
2. Dataset: Kaggle's Titanic Survival Dataset.
3. Accurcay obtained: 0.7886
4. Rank : Among top 27%

Kaggle's House Price Prediction Model

I would recommend every one to be a part of these extreme model which completely teaches us about the concept of encoding and data explorations and taking decision while exploiting info from the data. It included nearly 75 Variable which needed to be explored in the process of House-Selling Price Regression.

1. Model used:
Naive Bayes, Random Forest, XGBoost, AdaBoost, Support Vector, and an ensemble model of them
2. Dataset: Kaggle's House Price Dataset.
3. Mean Squared Accurcay obtained: 0.15066
4. Rank : Among top 57%

Kaggle's Real or Fake Tweets Competition

It was the second implementation podium for my earlier work wherein, as we use twitter now-a-days and share our messages which are now-a-days being explored to predict for the possible mishap which has occured or lying in future for us nd our loved one's, wherein I was grilled completely during implementing the model which was filled complicated tweets needed to be filttered out.

1. Model used:
Bag of Word Model along with different classifications models.
2. Dataset: Kaggle's House Price Dataset.
3. Mean Squared Accurcay obtained: 0.57462
4. Rank : Among top 65%