cassandra-mmm 0.1.10

Cassandra library

Cassandra

Cassandra's library for Marketing Mix Modeling

To best perform an mmm, you will need to perform these processes in order:

• Data Processing
• Data Analysis
• Trasformations
• Models
• Model Evaluation

Build your MMM

Create your settings variables (settings_csv, settings_variables, settings_models) as shown below:

• settings_csv

settings_csv = {
    'pathcsv': 'path/to/dataset.csv',
    'delimiter': ',',
    'subset_columns': ['ordine_data','revenue','transazioni','google_search_spent','google_performance_max_spent','google_display_spent','fb_retargeting_spent','fb_prospecting_spent','bing_spent','bing_organico','google_organico','referral','email_sessioni','cambio_eur_dollaro','ios_14','Offerte'],
    'rename': {'ordine_data':'data'},
    'data': 'data',
    'target': 'revenue'
}

• settings_variables

settings_variables = {
    'medias': ['google_search_spent','google_performance_max_spent','google_display_spent','fb_retargeting_spent','fb_prospecting_spent','bing_spent'],
    'organic': ['bing_organico', 'google_organico', 'referral', 'email_sessioni', 'cambio_eur_dollaro', 'ios_14',
               'Offerte', 'lun', 'mar', 'mer', 'gio', 'ven', 'sab', 'dom', 'month', 'year'],
    'seasonal_categorical': ['month', 'year', 'lun', 'mar', 'mer', 'gio', 'ven', 'sab', 'dom'],
    'general_categorical': ['ios_14', 'Offerte'],
    'general_numeric': ['bing_organico', 'google_organico', 'referral', 'email_sessioni', 'cambio_eur_dollaro']
}

• settings_models

settings_models = {
        "model": 'nevergrad',
        "linear_metric": ['accuracy', 'nrmse', 'mape'],
        "linear_coeffs": True,
        "ols_constant": True,
        "log_linear_metric": ['accuracy', 'nrmse', 'mape'],
        "log_linear_coeffs": True,
        "ridge_metric": ['accuracy', 'nrmse', 'mape'],
        "deep_learning_metric": ['accuracy', 'nrmse', 'mape'],
        "nevergrad_metric": ['accuracy', 'nrmse', 'mape'],
        'nevergrad_variables': {
            'google_search_spent': {
                'lower':0,
                'upper':3
            },
            'google_performance_max_spent': {
                'lower':0,
                'upper':6
            },
            'google_display_spent': {
                'lower':0,
            }
        },
        'nevergrad_budget': 2500
    }

Run your MMM

main.build(settings_csv, settings_variables, settings_models)