In this project, I conducted a comprehensive analysis of the Kaggle dataset “House Prices – Advanced Regression Techniques,” featuring 1460 observations of homes in Ames, Iowa. With 79 explanatory variables, I used regression models to predict house prices based on factors such as physical characteristics (e.g., size, number of rooms, amenities), location, and temporal factors like the year built and remodeled.

Key areas of analysis include:

• Impact of Physical Characteristics: Exploring how home size, number of bedrooms, bathrooms, and amenities affect pricing.
• Location-Based Insights: Comparing neighborhoods and examining the influence of proximity to streets on property values.
• Temporal Factors: Analyzing the effect of build and remodel years on house prices to guide investment decisions.

This project provides actionable insights for real estate developers, home buyers, and community planners, helping them make data-driven decisions based on property features and market trends.

Developed an LSTM model using Python to help forecast United States COVID-19 cases and achieved a loss of 6.755e-06.ignificantly enhancing the reliability and precision of epidemiological forecasting, which is critical for strategic
public health planning and response efforts. Extracted COVID-19 case data from Kaggle in the last six months and performed data preprocessing steps using Pandas and Scikit-learn. Also Visualized the COVID cases forecast using Numpy and Seaborn

In this project, I developed a program to project 3D skeletal data onto 2D images using a dataset of human poses. The dataset consisted of 3D joint coordinates along with camera position and focal length information. My task involved:

• Reading the 3D joint positions and camera data.
• Computing the correct camera orientation based on the provided camera positions.
• Projecting the 3D skeleton onto a 2D plane to generate a visual representation of the human pose.

The project produced 2D projections of the 3D skeletons for 20 images, visualized on a blank background and compiled into a table for validation. This solution demonstrates the application of geometric transformations to project 3D data into 2D space while keeping a high accuracy on joints tracking.

This project focused on building a machine learning model to predict the duration of taxi trips in New York City. The dataset included features such as pickup time, number of passengers, and geo-coordinates, which provided a comprehensive foundation for prediction. My task involved:

• Data Exploration and Feature Engineering: Conducting an in-depth analysis of the dataset to understand relationships between variables, clean the data, and engineer new features to enhance the predictive power of the model.
• Neural Network Modeling: Developing and training a neural network model capable of capturing the complex, non-linear interactions between various features in the data. The choice of neural networks allowed for greater flexibility in handling high-dimensional data and uncovering hidden patterns.
• Bayesian Optimization for Hyperparameter Tuning: Applying Bayesian Optimization to fine-tune the neural network’s hyperparameters. This method efficiently explored the hyperparameter space, reducing computation time and improving the model’s overall accuracy compared to traditional methods like grid search.
• Model Evaluation: Evaluating the performance of the model based on specific metrics to ensure it provided accurate predictions across different trip durations, with a focus on minimizing error and ensuring consistency.

The project successfully produced a model capable of accurately predicting taxi trip durations, showcasing the ability of neural networks to handle complex data relationships. The use of Bayesian Optimization played a key role in refining the model’s performance. This approach demonstrates how advanced machine learning methods can be effectively applied to solve practical problems in transportation, delivering both accuracy and efficiency in prediction.