Here are some of the projects I have worked on.
○ Achieved a 91% accuracy, 89% recall rate in forecasting ethnic after extensive data preprocessing.
○ Handled class imbalance by applying SMOTE method, increasing macro average recall score by 7%.
○ Received the best score with logistic regression model compared to SVC, random forest, and AdaBoost.
○ Showed feature impact on the final model using the SHAP method.
○ Conducted exploratory data analysis (EDA) and outlier detection to ensure data quality and reliability.
○ Performed RFM analysis to segment customers into low, middle, and top groups.
○ Applied K-Means clustering and visualized the results in an understandable graph.
○ Developed an artificial neural network (ANN) architecture, prevented overfitting, and handled class imbalance using the SMOTE method.
○ Increased accuracy score by 24% after handling outliers without overfitting in the final architecture.
○ Analyzed job trends across different data roles.
○ Employed statistical methods such as Kruskal-Wallis and Dunn tests to compare mean salaries.
○ Used chi-squared tests to examine relationships between categorical variables and linear regression to model the impact of various factors on salary.
○ Performed data cleaning, outlier detection, and exploratory data analysis (EDA).
○ Analyzed sales trends, customer behavior, and product performance.
○ Visualized results using interactive Plotly visualizations.
○ Performed sentiment analysis on Amazon product reviews to classify them as positive ,neutral or negative.
○ Applied text preprocessing including lemmatization, stopword removal, and frequency filtering.
○ Visualized word frequencies and created word clouds for key insights.
○ Compared Logistic Regression and Random Forest models, achieving strong classification performance.