In this chapter you will learn how to construct the foundations of your base table, namely the population and the target.

The basetable timeline

Timeline violations

Available data

Timeline violation

The population

Select the relevant population

A timeline compliant population

Removing duplicate objects

The target

Calculate an event target

Calculate an aggregated target

Crucial base table concepts

You will learn how to add variables to the base table that you can use to predict the target.

Adding fixed variables

Selecting the right value

Adding age

Adding the donor segment

Adding living place

Adding aggregated variables

Selecting the appropriate date

Maximum value last year

Recency of donations

Adding evolutions

Ratio of last month's and last year's average

Absolute difference between two years

Using evolution variables

Performance of evolution variables

Meaning of evolution

Creating variables

Once you derived variables from the raw data, it is time to clean the data and prepare it for modeling. In this Chapter we discuss the steps that need to be taken to make your data modeling-ready.

Creare variabili fittizie

Creare una dummy da una variabile con due categorie

Creare dummies da una variabile con molte categorie

Valori mancanti

Come sostituire i valori mancanti

Creare una variabile fittizia per i valori mancanti

Sostituisci i valori mancanti con il valore mediano

Sostituisci i valori mancanti con un valore fisso

Gestire i valori anomali

Influenza degli outlier sui modelli predittivi

Gestire gli outlier con la winsorizzazione

Gestire gli outlier con la deviazione standard

Trasformazioni

Interazioni

Trasformazione con radice quadrata

Aggiungere interazioni al basetable

Data preparation

In some cases, the target or variables change heavily with the seasons. You will learn how you can deal with seasonality by adding different snapshots to the base table.

Seasonality

Seasonality or not

Detecting seasonality

The effect of seasonality

Using multiple snapshots

Target values

Calculating snapshot targets

Calculating aggregated variables

Stacking basetables

The timegap

Events during the timegap

Calculating aggregated variables with timegap

Adding age with timegap

Congratulations

Advanced base table concepts

Donor IDs

Basetable with countries and age

Basetable used in Ex 2.13

Living place of donors

Donations

Costruire buoni modelli è possibile solo se parti da una base table solida. In questo corso imparerai a costruire una buona base table, a creare variabili e a preparare i dati per il modeling. Concludiamo con argomenti avanzati sull’argomento.

Introduction to Predictive Analytics in Python

Impara a preparare e organizzare i dati per l'analisi predittiva con Python e concetti avanzati.

Analisi predittiva intermedia in Python

Scopri come preparare e organizzare i tuoi dati per l'analisi predittiva.

Influenza degli outlier sui modelli predittivi

Analisi predittiva intermedia in Python

Esercizio pratico interattivo