MLDL-Workshop

For Colab users

1. Open Notebook

   https://colab.research.google.com/

   File -> Open Notebook -> Github -> Paste:

   https://github.com/J0nasSeng/MLDL-Workshop/tree/main/notebooks

   Click 🔍 -> data-cleaning.ipynb

2. Upload Data

   local: download and zip the folder data to data.zip

   colab: click 📁.. -> click upload -> select data.zip -> upload

   unzip in colab: add the following line before loading the data and run the cell

   !unzip ../data.zip -d ../
   

3. (possible) Typos in the notebooks

   3.1 data-cleaning.ipynb cell 40 line 2

   if credit_data[c].dtype == 'object' and credit_data[c].name != 'Credit_Mix' and credit_data[c].name != 'Payment_of_Min_Amount' and credit_data[c].name != 'Credit_Score':
   

   3.2 random-forests.ipynb Section Gradient Boosting, cell 2

   xgb.XGBClassifier(....  , num_class=len(y_train.unique())
   

   3.3 glms.ipynb Before classification

   print(X.columns[1])
   print(X.columns[2])
   print(X.columns[9])
   

   3.4 probabilistic-circuits.ipynb before import

   !git clone https://github.com/SPFlow/SPFlow.git
   # Fix the import error in SPFlow by replacing NINF with float('-inf')
   !sed -i "s/from scipy import NINF/NINF = float('-inf')/" /content/SPFlow/src/spn/algorithms/splitting/PoissonStabilityTest.py
   %cd SPFlow/src
   !bash create_pip_dist.sh
   !pip install dist/spflow-0.0.40-py3-none-any.whl --no-deps
   %cd ../..
   

   3.5 hyperparameter-optimization.ipynb before import

   !pip install scikit-optimize