Meghana Sulake

Domain: Research work from IIT Bombay

Objective: Develop a Machine Learning regressor model which can predict amount of Heat Release Rate (HRR) value.

• Demonstrated the ability to fetch, process and leverage data to generate useful predictions.
• Understanding the data and perform EDA.
• Missing value check, outlier detection, scaling and normalization, Univariate and bivariate data analysis using jointplot & pairplot.
• Visualized a heatmap to understand correlation between all features.
• Trained a Supervised Learning Regression base model using Linear Regressor, SVR, KNN, Decision Tree & Random forest.
• Tried various hyper-parameters to improve performance of the selected model.

Domain: Telecom

Objective: Build a model that will help telecom company to identify the potential customers who have a higher probability to churn.

• Data Understanding & Exploration.
• Performed data Cleaning & Analysis.
• Visualized percentage distribution in the pie-chart.
• Trained a model using Decision Tree, Random Forest, Adaboost, Bagging, Gradient Boost and show best performing model.
• Improve performance of the model as much as possible by tuning different hyper-parameters and show best performing parameters along with train and test performance.

Domain: Automobile

Objective: Build a model to classify a given silhouette as one of three types of vehicles, using a set of features extracted from the silhouette.

• Performed Data Extraction, cleaning and Analysis.
• Apply PCA on the data with 10 components.
• Visualize Cumulative Variance Explained with Number of Components and Select Minimum Components with 90% or above variance explained.
• Developing a base Classification model using SVM which will classify the type of vehicle

Domain: Smartphone, Electronics

Objective: Build a Recommendation System model using popularity based and collaborative filtering methods to recommend mobile phones to a user which are most popular and personalized respectively.

• Understand the data and perform EDA.
• Build a popularity-based model and recommend top 5 mobile phones.
• Used surprise library and build a collaborative filtering model using SVD and kNNWithMeans.