RAJU BHATTARAI

Motivated data professional with strong foundations in Mathematics, Data Science, machine learning, and data analysis. Experienced in building classification models, dimensionality reduction, and data visualization using Python, SQL, and MATLAB. Skilled with real-world datasets such as MNIST, dedicated to transforming data into meaningful insights to support decision-making.

• Neural Network for MNIST Classification (Python, NumPy)

Implemented a forward/backward propagation neural network from scratch using NumPy. Designed architecture, activation functions, and gradient descent manually. Trained on MNIST digits and achieved 87.1% accuracy while evaluating loss curves and tuning hyperparameters.

• KNN for Digit Recognition (Python, scikit‑learn)

Built a K‑Nearest Neighbors classifier for MNIST. Compared k‑values, confusion matrices, evaluated accuracy‑speed tradeoffs, achieving 94.7% accuracy with k=1.

• PCA/SVD Dimensionality Reduction (MATLAB)

Applied PCA and SVD to denoise high‑dimensional time‑series data. Retained major variance, reconstructed cleaner signals, and visualized the impact of singular values on dimensionality reduction.

• Isomap for Nonlinear Dimensionality Reduction (MATLAB)

Computed geodesic distances, constructed neighborhood graphs for Swiss roll dataset, used shortest‑path matrices and MDS to extract intrinsic 2D geometry and analyzed eigenvalue spectra