Predictive Analytics Techniques: Theory and Applications in Finance

Abstract

This chapter presents several models associated with predictive analysis across disciplines. While the models are presented to appeal to finance professionals and learners, they were chosen because of their wide use across disciplines. The chapter covers five models: logistic regression, time series analysis, decision trees, multiple linear regression, and RFM (Recency, Frequency, Monetary) Segmentation with k-means. The models are presented with moderate mathematical depth and with an emphasis on building a working software implementation in the R programming language. The theoretical justifications are weaved with the software creation, under the premise that seeing the model work provides the necessary encouragement to learn the theory (in a different book perhaps). Some datasets are accessed from public repositories, while other are synthetic and have been created specifically for this project. Each model is built to be used in finance as well as in any discipline in which similar questions are asked of the data.

Key Terms and Definitions

ARIMA :

Auto-regressive Integrated Moving Average, is one of the two most widely used approaches to time series forecasting. It aims to describe autocorrelations in the data as predictors of future values.

AUC curve :

Area Under the ROC Curve.

Exponential smoothing :

One of the two most widely used approaches to time series forecasting. It is based on a description of the trend and seasonality in the data.

Gain and Lift chart :

A chart used to evaluate the performance of a (classification) model. It shows the difference between making predictions using a model and without the model.

k -means :

An unsupervised clustering algorithm, where k is the number of clusters, set by the user.

Logistic regression :

A statistical model that uses the logistic function (sigmoid shaped) to model a binary dependent variable (outcome).

Non-Constant Variance (NCV) test :

Computes a score test of the hypothesis of constant error variance against the alternative that the error variance changes with the level of the response (fitted values), or with a linear combination of predictors.

Odd ratio :

A measure of association between an independent variable and an outcome. It represents the odds that an outcome will occur given a particular event, compared to the odds of the outcome occurring in the absence of that event.

RFM :

Recency, frequency, monetary value is a marketing analysis tool, using measures to identify best customers for a business.

ROC curve :

Receiver Operating Characteristic curve. It visualizes the ratio between TPR (True Positive Rate) and FPR (False Positive Rate).

Seasonal Variation :

A component of a time series which is defined as the repetitive and predictable movement around the trend line in 1one year or less.

Time Series :

A set of points collected over a period of time.

Tree pruning :

A process to reduce the size of a decision tree, by removing noncritical subtrees.