Standardizing continuous numerical features

continuous_columns = data.select_dtypes(include=['float64']).columns.tolist()

• scaler = StandardScaler()
• scaled_features = scaler.fit_transform(data[continuous_columns])

Converting to a DataFrame

scaled_df = pd.DataFrame(scaled_features, columns=scaler.get_feature_names_out(continuous_columns))

Combining with the original dataset

scaled_data = pd.concat([data.drop(columns=continuous_columns), scaled_df], axis=1)

Frequently Used Codes.

Identifying categorical columns

• categorical_columns = scaled_data.select_dtypes(include=['object']).columns.tolist()
• categorical_columns.remove('NObeyesdad') # Exclude target column

Applying one-hot encoding

• encoder = OneHotEncoder(sparse_output=False, drop='first')
• encoded_features = encoder.fit_transform(scaled_data[categorical_columns])

Converting to a DataFrame

encoded_df = pd.DataFrame(encoded_features, columns=encoder.get_feature_names_out(categorical_columns))

Combining with the original dataset

prepped_data = pd.concat([scaled_data.drop(columns=categorical_columns), encoded_df], axis=1)

Encoding the target variable

• prepped_data['NObeyesdad'] = prepped_data['NObeyesdad'].astype('category').cat.codes
• prepped_data.head()

• I uses these codes alot and this repository works like a library.
• Codes that are essential to train Neural Networks and use Google Colab are listed.
• Code used for outlier treatment.
• Basic python data visualization code.