From Prediction to Causality in Data Science

1. Introduction

1.1.1 Evolution of Data Science: From Description to Prescription

There are three significant phases of data science development. The first phase was on descriptive analytics where statistical software were employed to summarize past data and report what has already occurred. This involves standard examination of economic indicators like rates of growth, inflation and employment rates in economics. The second phase presented predictive analytics, which was motivated by progress in machine learning and massive data availability. In this case, models are conditioned to predict the future, which is market tendencies, demand, or macroeconomical variables.

In more recent times the emphasis has moved on to prescriptive analytics besides describing and predicting outcomes it tries to aid in decision-making. Prescriptive methods attempt to prescribe action by integrating evidence-based models and decision goals. This shift is indicative of a wider trend that data science must not only analyze the world, but also assist in transforming it in a favorable manner. Here, the role of data science is not only growing out of a passive analysis tool but also an active part of a decision-making system, particularly in the economic and public policy world (Lo and Pachamanova, 2023) [1]. Figure 1.1 is an image that represents the evolution of data science paradigms between descriptive and causal data science.