Skip to content

MAC_INSTALL_FIX.md

Latest commit

 

History

History
164 lines (117 loc) · 3.21 KB

MAC_INSTALL_FIX.md

File metadata and controls

164 lines (117 loc) · 3.21 KB
Raw

Mac Installation Fix for Python 3.13

The Problem

Python 3.13 is very new (released Oct 2024) and some packages don't have pre-built wheels yet, causing compilation errors.

Solution: Install in Stages

Stage 1: Core Dependencies (Fast)

cd backend

# Install minimal requirements first
pip install --upgrade pip
pip install numpy scipy
pip install scikit-learn
pip install Flask Flask-CORS
pip install textblob nltk
pip install matplotlib seaborn
pip install pyyaml python-dotenv

Stage 2: Computer Vision (May take time)

# OpenCV - may need to compile
pip install opencv-python
pip install opencv-contrib-python

# Alternative: Use brew-installed OpenCV
# brew install opencv

Stage 3: Deep Learning (Optional, takes longest)

# PyTorch - large download
pip install torch torchvision --index-url https://download.pytorch.org/whl/cpu

# Transformers
pip install transformers datasets accelerate

# TensorBoard
pip install tensorboard

# SHAP
pip install shap

Stage 4: MLOps (Optional)

pip install mlflow
pip install textstat

Quick Test After Stage 1

# Test feature extraction
python feature_extraction.py

# Test experiment tracker (no dependencies needed)
python experiments/experiment_tracker.py

Alternative: Use Python 3.11

If you continue having issues, use Python 3.11 (most stable):

# Install Python 3.11 using pyenv
brew install pyenv
pyenv install 3.11.6
pyenv local 3.11.6

# Create new venv
python -m venv venv
source venv/bin/activate

# Install all requirements
pip install -r requirements.txt

Skip Heavy Dependencies

If you just want to see the code working:

# Install only what's needed for basic tests
pip install -r requirements-minimal.txt

# Download NLTK data
python -c "import nltk; nltk.download('brown'); nltk.download('punkt')"

What Works Without Full Installation

These components work with minimal dependencies:

✅ Experiment Tracker - No ML dependencies needed

python experiments/experiment_tracker.py

✅ Configuration Management - Just needs PyYAML

pip install pyyaml
cat config/experiment_config.yaml

✅ Documentation - Read without running

open FEATURES.md
open CV_ANALYSIS.md
open MLOPS.md

What Needs Full Installation

These need all dependencies:

❌ Feature Analysis (needs sklearn, shap, matplotlib) ❌ Computer Vision (needs opencv) ❌ Transformer Training (needs torch, transformers) ❌ Model Comparison (needs sklearn, transformers)

Recommended Approach for Demo

  1. Install minimal requirements (5 min)
  2. Run experiment tracker (works immediately)
  3. Read documentation (no installation needed)
  4. Show code quality (no need to run)

This is sufficient for portfolio/interview demos!

If Compilation Errors Persist

# Install Xcode Command Line Tools
xcode-select --install

# Update Homebrew
brew update

# Install OpenMP support
brew install libomp

# Try again
pip install scikit-learn

Current Status Check

# See what's installed
pip list

# Test imports
python -c "import numpy; print('numpy OK')"
python -c "import sklearn; print('sklearn OK')"
python -c "import flask; print('flask OK')"