Learn Data Skills
Beta
Mina Mehdinia

Mina Mehdinia

Self-employed

Self-employed | Chicago

Technologies

Data Scientist & Analyst | Machine Learning & Gen AI Enthusiast

My Work

Take a look at my latest work.

article

Fine Tunning BERT Model for Amazon Product Review and Deploying it into Hugging Face Model Hub: | by Mina Mehdinia | Medium

Python
article

Fine-tuned BERT Embeddings and T-SNE Visualization | by Mina Mehdinia | Nov, 2023 | Medium

Python
article

Automating Review Evaluation with OpenAI’s GPT-3.5 | by Mina Mehdinia | Medium

Python
article

Fine Tunning BERT Model for Amazon Product Review and Deploying it into Hugging Face Model Hub: | by Mina Mehdinia | Medium

Python

DataCamp Course Completion

Take a look at all the courses I’ve completed on DataCamp.

Courses

Introduction to Python
Introduction to SQL
Intermediate SQL
Joining Data in SQL

My Work Experience

Where I've interned and worked during my career.

Self-employed | Jun 2023 - Present

Freelance Data Scientist

• Fine-tuned BERT: Enhanced and automated product rating predictions on Amazon by fine-tuning a 110M params BERT model using Hugging Face’s transformers library, achieving 52% improvement over ChatGPT. Deployed the model to the Hugging Face Model Hub for broader accessibility (Try it)(https://huggingface.co/minamhd/BERT_Amazon_Review). • Prompt Engineering with GPT: Leveraged OpenAI’s GPT-3.5 to elevate product review scoring and analysis, achieving a very low error (MAE of 0.56) in sentiment accuracy and led innovations in zero-shot and few-shot prompting techniques. • Naive Bayes Classifier: Spearheaded the development of a Naive Bayes classifier as a baseline model, achieving 67% prediction accuracy for Amazon product ratings based on customer reviews.
Show More

Portland State University | Sep 2021 - Jun 2023

Data Analyst Research Assistant

• Loan Default Prediction: Streamlined predictive modeling using dimensionality reduction, achieving 81.4% accuracy in loan default prediction by employing advanced algorithms. • Impact Assessment: Executed in-depth exploratory data analysis, unveiling crucial insights about COVID’s effect on transgender women’s healthcare. • Empowered Decision Making: Assessed Klamath River data to unveil key factors impacting thermal stratification, resulting in 87.68% better strategies for river management. • EDA and predictive modeling: Designed and executed a web scraping strategy in R, capturing data from the Cherry Blossom 10-mile race (1973-2022). Streamlined data preparation in R and employed a Linear Mixed Effects (LME) model, uncovering a notable link between age and running speed influenced by physical fitness.

Summer Institute in Bio-statistics and Data Science, UCI | Jul 2022 - Aug 2022

Data Science Intern

• Statistial Modeling: Spearheaded biostatistical modeling techniques, achieving a 68% correlation between stress and biophysical, contextual, and demographic features. Improved model interpretability by 40% through advanced linear modeling.

Portland State University | Jan 2022 - Jun 2022

Machine Learning Research Assistant

• Object Detection: Conceived and executed a deep learning system using PyTorch that accurately computed the total value of US coins under varied lighting and backgrounds achieving 97.5% precision. • Recommendation System: Led the development of K-means clustering and KNN regression models to recommend food items, achieving a 93.3% accuracy in food categorization recommendations.

My Education

Take a look at my formal education

Bachelor's degree in Data SciencePortland State University | 2023

About Me

Mina Mehdinia

Mina hasn't filled in a bio text

Powered by

  • Work
  • Courses
  • Experience
  • Education
  • About Me
  • Create Your Data Portfolio for Free