1. Data cleaning, visualization, correlation, transformation

• Clean errors of the columns with a function
• Create random column (to be used in the feature selection)
• Column transformation

download clean csv without filling csv

• Fill missing values (ffill)

download clean_no_nan to csv

2. ML Modeling

• Encoding data

download encoded data to csv

• define X and y
• Balance data (undersample NearMiss)
• Split data (train_test_split)
• Scale data (StandardScaler)
• Run each model
• Short description about each model
  • In RandomForest do the feature selection
  • Run models with only selected features (probably will do it in the pipeline)
• Visualize model results
• Hyperparameter tuning for each model with GridSearch
• Run each model with best parameters
• Visualize model results

3. ML Modeling (pipeline and deep learning)

• Create pipeline and try all models models
• Try with different scalers
• Try with oversampling (SMOTE)
• Other changes (different percentage of train and test..)
• Cross-validation with and without kfold
• Run one deep learning model

---

• 1st part - Sosa (with Isaak if has the function finished)
• 2nd part - Kristina with Sosa (Sosa taking care of visualizations)
• 3rd part - Pau and Jota
• Readme - Isaak
• Presentation ?