Hiba Lafkir

Data Science Student

Passionate about AI and big data tools, I enjoy building machine learning solutions that solve real-world problems — from predicting real estate prices in Casablanca to detecting skin lesions with deep learning. Always learning, always building.

Python Machine Learning Deep Learning Computer Vision scikit-learn PyTorch Big Data

Casablanca House Price Prediction Regression

Predicts apartment and house prices in Casablanca from scraped real estate listings. Built a full ML pipeline with regex-based feature extraction, neighborhood price signals, and a Stacking Regressor combining Linear Regression, Ridge, and Decision Tree models. Includes an interactive Streamlit app for live predictions.

Stacking

Best model

MAE

Optimized metric

5-fold CV

Validation strategy

pandas scikit-learn Streamlit GridSearchCV joblib regex

View on GitHub →

Skin Lesion Segmentation — Mask R-CNN Segmentation

Fine-tuned a Mask R-CNN model with a ResNet-50 FPN backbone on the ISIC 2016 challenge dataset to automatically segment skin lesions from dermoscopic images. Generates pixel-level masks that could support early melanoma detection in clinical workflows.

85.9%

Mean IoU

92.6%

Precision

95.6%

Accuracy

PyTorch torchvision Mask R-CNN ResNet-50 FPN ISIC 2016 OpenCV

View on GitHub →