Descriptive, Predictive, and Prescriptive Analysis represent the three stages of analytical maturity.

Most organizations are stuck in the first stage (Descriptive)—looking at dashboards to see what happened yesterday. The real value lies in moving up the curve: predicting what will happen and, crucially, using algorithms to decide what to do about it.

Here is the detailed breakdown of the three analysis tiers, the algorithms involved, and the toolset, followed by the downloadable Word file.

1. The Analytics Maturity Model

• Descriptive (Hindsight): "We lost 15 customers last month."
• Tools: SQL, PowerBI, Excel.
• Predictive (Foresight): "We are likely to lose these 5 specific customers next month."
• Tools: Python (Scikit-Learn), R, AutoML.
• Prescriptive (Optimization): "Offer these 5 customers a 10% discount to prevent them from leaving."
• Tools: Linear Programming, Reinforcement Learning, Simulation.

2. Tier 1: Descriptive Analysis (The Foundation)

This is about summarizing history to understand the present.

• Techniques:
• Aggregation: Calculating Mean, Median, Mode, and Standard Deviation.
• Clustering (Segmentation): Grouping customers based on behavior (e.g., "High Spenders," "Bargain Hunters") using K-Means Clustering.
• Data Visualization: Using histograms and scatter plots to spot trends and outliers.

3. Tier 2: Predictive Analysis (The Crystal Ball)

This uses historical data to train models that forecast the future.

A. Regression (Forecasting Numbers)

• Use Case: "What will our sales be in Q4?"
• Algorithm: Linear Regression or XGBoost.
• Input: Past sales, marketing spend, seasonality.
• Output: A specific number (e.g., "$1.5M").

B. Classification (Predicting Outcomes)

• Use Case: "Is this transaction fraudulent?"
• Algorithm: Logistic Regression or Random Forest.
• Input: Transaction amount, location, user device.
• Output: A probability (e.g., "98% Fraud").

4. Tier 3: Prescriptive Analysis ( The Decision Maker)

This is where analysis meets action. It tells you the optimal path.

• Optimization (Linear Programming):
• Scenario: You have 5 trucks and 50 deliveries.
• Goal: Minimize fuel cost.
• Constraints: Trucks have max weight; drivers have max hours.
• Result: The algorithm outputs the exact route every driver should take.
• Simulation (Monte Carlo):
• Scenario: "What if we raise prices by 5%?"
• Method: The computer runs the scenario 10,000 times with random variables (market crash, competitor launch) to give you a probability distribution of the outcome.

5. Key Applications & Tools

Category	Tool	Usage
Descriptive	PowerBI / Tableau	Interactive dashboards to visualize "What happened."
Predictive	Scikit-Learn	The standard Python library for Machine Learning (Regression, Classification).
	Prophet	Facebook's tool for easy, accurate time-series forecasting.
Prescriptive	Gurobi / CPLEX	Heavy-duty solvers for optimization problems (Supply Chain, Scheduling).
AutoML	DataRobot / H2O	Automates the model building process. You upload data; it picks the best algorithm.