In the last session, we discussed how Machines learn i.e. Supervised, Unsupervised, and Reinforcement Learning. More and more companies are building products leveraging these learning techniques. But companies are not only leveraging data to teach machines but also leveraging data to learn more about the past, and the future, and make business recommendations. This is what is called Advanced Analytics.

AI is about using data to make machines smart while Advanced Analytics is about using data to make humans smart

‍

Advanced Analytics consists of three important pillars - Each building on the other but answering three different questions.

Descriptive Analytics:

• Goal: Descriptive analytics focuses on summarizing historical data and identifying patterns or trends to understand “What has happened in the past?”.
• Methodologies: Descriptive analytics techniques include data summarization, visualization, and exploratory data analysis (EDA). These techniques help in organizing and presenting data in a meaningful way to gain insights.
• Applications: Descriptive analytics is commonly used for reporting, dashboarding, and data visualization to monitor business performance, track KPIs, and identify trends. When you walk into your company and see all the Tableau, Looker, and Google Analytics Dashboards, It is all part of Descriptive Analytics.
• Example: Suppose you are running an ice cream store. Your data analyst looked at historical data and told you ‘Sales go up every Summer’. This is descriptive analytics.

Predictive Analytics:

• Goal: Predictive analytics aims to forecast future outcomes or trends based on historical data patterns. It seeks to answer questions like "What will happen in the future?"
• Methodologies: Predictive analytics techniques involve building statistical or machine learning models on historical data to make predictions. Common techniques include regression analysis, time series forecasting, and machine learning algorithms such as decision trees, random forests, and neural networks. We will talk about these models in the incoming modules.
• Applications: Predictive analytics is used for demand forecasting, risk assessment, sales forecasting, customer churn prediction, and other predictive modeling tasks.
• Example: In your ice cream store, the data analytics look at historical data and realize Ice cream sales go up every summer. Using this to predict that the sales will go up the next summer also is predictive analytics. You are predicting what will happen.

Prescriptive Analytics:

• Goal: Prescriptive analytics goes beyond predicting future outcomes to provide recommendations on what actions to take to achieve desired outcomes. It seeks to answer questions like "What should we do?"
• Methodologies: Prescriptive analytics techniques involve optimization, simulation, and decision analysis. These techniques help in identifying the best course of action based on predictive models, business constraints, and objectives.
• Applications: Prescriptive analytics is used for resource allocation, supply chain optimization, pricing optimization, treatment planning in healthcare, and other decision-making scenarios where optimal decisions need to be made.
• Example: Back to your ice cream store. The data analyst looked at the historical data and realized that when prices go down by 10%, sales increase by 25%. So he can recommend reducing the price by 10% to increase sales. This is prescriptive analytics helping answer what should you do to achieve your business goals.

‍

How Advanced Analytics is used in Organizations

Most companies are interested in answering questions - what will happen and what should happen more than what happened. Why?

It is because as interesting as history is, it has passed. It is of no relevance in the future. If that is the case, then why do companies or data science teams spend most time on Descriptive Analytics?

Two reasons:

1. History repeats itself - In the absence of any crystal ball, history is the best available predictor of the future. So studying history is highly important.
2. You cannot do Prescriptive Analytics or Predictive Analytics without doing Descriptive Analytics. However, this reason is greatly tied to the first reason.

‍

‍

So while companies want to spend more time answering what should happen i.e. Prescriptive Analytics, they spend more time answering What happened i.e. Descriptive Analytics.