The Project | About Us | Contribute | Donations | License

HOME

Prescriptive analysis is a powerful technique used in business analysis to recommend actions you can take to affect desired outcomes. It goes beyond descriptive and predictive analysis by not only forecasting future scenarios but also suggesting the best courses of action to achieve specific goals. This module will cover the fundamentals of prescriptive analysis, including its definition, methods, tools, and practical applications.

What is Prescriptive Analysis?

Prescriptive analysis involves the use of data, algorithms, and machine learning to determine the best course of action for a given situation. It answers the question, "What should we do?" by providing actionable recommendations based on data-driven insights.

Key Components of Prescriptive Analysis:

Data Collection: Gathering relevant data from various sources.
Data Processing: Cleaning and organizing the data for analysis.
Modeling: Using algorithms and models to analyze the data.
Simulation: Running simulations to test different scenarios.
Optimization: Finding the best solution among various alternatives.
Implementation: Applying the recommended actions.

Methods of Prescriptive Analysis

Optimization Models

Optimization models are mathematical models that help in finding the best solution from a set of feasible solutions. These models are used to maximize or minimize an objective function, such as profit, cost, or efficiency.

Example: Linear Programming

from scipy.optimize import linprog

# Objective function coefficients (maximize profit)
c = [-20, -30]

# Inequality constraints (resource limitations)
A = [[1, 2], [3, 1]]
b = [40, 30]

# Bounds for variables (non-negative)
x0_bounds = (0, None)
x1_bounds = (0, None)

# Solve the linear programming problem
result = linprog(c, A_ub=A, b_ub=b, bounds=[x0_bounds, x1_bounds], method='simplex')

print('Optimal value:', -result.fun, '\nX:', result.x)

In this example, we use linear programming to maximize profit given certain resource constraints.

Simulation Models

Simulation models are used to mimic the operation of a real-world process or system over time. They help in understanding the impact of different variables and scenarios.

Example: Monte Carlo Simulation

import numpy as np

# Number of simulations
num_simulations = 10000

# Simulate random variables (e.g., sales and costs)
sales = np.random.normal(100, 10, num_simulations)
costs = np.random.normal(50, 5, num_simulations)

# Calculate profit
profit = sales - costs

# Analyze the results
mean_profit = np.mean(profit)
std_profit = np.std(profit)

print('Mean Profit:', mean_profit)
print('Profit Standard Deviation:', std_profit)

This example demonstrates how Monte Carlo simulation can be used to estimate the distribution of profit based on random variables.

Decision Analysis

Decision analysis involves evaluating and comparing different decision options using decision trees, payoff matrices, and other tools.

Example: Decision Tree Analysis

from sklearn.tree import DecisionTreeClassifier
from sklearn import datasets

# Load dataset
iris = datasets.load_iris()
X = iris.data
y = iris.target

# Train decision tree classifier
clf = DecisionTreeClassifier()
clf.fit(X, y)

# Predict class for a new sample
sample = [[5.1, 3.5, 1.4, 0.2]]
prediction = clf.predict(sample)

print('Predicted class:', prediction)

In this example, a decision tree classifier is used to predict the class of a new sample based on the Iris dataset.

Tools for Prescriptive Analysis

Optimization Software

IBM ILOG CPLEX: A high-performance mathematical programming solver.
Gurobi: An optimization solver for linear programming, quadratic programming, and more.

Simulation Software

AnyLogic: A simulation software for business processes, supply chains, and more.
Arena: A discrete event simulation and automation software.

Decision Analysis Tools

TreeAge Pro: A tool for building decision trees and performing sensitivity analysis.
Analytica: A visual software for creating, analyzing, and communicating decision models.

Practical Applications of Prescriptive Analysis

Supply Chain Optimization

Prescriptive analysis can optimize inventory levels, transportation routes, and production schedules to minimize costs and improve efficiency.

Marketing Campaign Optimization

By analyzing customer data, prescriptive analysis can recommend the best marketing strategies to maximize ROI.

Financial Planning

Prescriptive analysis helps in creating optimal investment portfolios, budgeting, and financial forecasting.

Exercise: Optimizing a Marketing Campaign

Problem Statement

A company wants to optimize its marketing campaign to maximize customer acquisition while staying within a budget of $50,000. The company has two marketing channels: online ads and TV commercials. Each online ad costs $500 and is expected to acquire 30 customers, while each TV commercial costs $5,000 and is expected to acquire 400 customers.

Task

Formulate the optimization problem.
Use linear programming to find the optimal number of online ads and TV commercials.

Solution

from scipy.optimize import linprog

# Objective function coefficients (maximize customer acquisition)
c = [-30, -400]

# Inequality constraints (budget limitation)
A = [[500, 5000]]
b = [50000]

# Bounds for variables (non-negative)
x0_bounds = (0, None)
x1_bounds = (0, None)

# Solve the linear programming problem
result = linprog(c, A_ub=A, b_ub=b, bounds=[x0_bounds, x1_bounds], method='simplex')

print('Optimal number of online ads:', result.x[0])
print('Optimal number of TV commercials:', result.x[1])
print('Maximum customer acquisition:', -result.fun)

Explanation

The objective function coefficients represent the number of customers acquired per online ad and TV commercial.
The inequality constraint ensures the total cost does not exceed the budget.
The solution provides the optimal number of online ads and TV commercials to maximize customer acquisition.

Common Mistakes and Tips

Common Mistakes

Ignoring Constraints: Ensure all constraints are properly defined and included in the model.
Overfitting Models: Avoid overly complex models that may not generalize well to new data.
Misinterpreting Results: Carefully interpret the results and consider the practical implications of the recommendations.

Tips

Validate Models: Regularly validate and update models with new data to ensure accuracy.
Collaborate with Stakeholders: Work closely with stakeholders to understand their needs and constraints.
Use Visualization: Visualize data and results to communicate findings effectively.

Conclusion

Prescriptive analysis is a crucial tool in business analysis that helps organizations make informed decisions and optimize their operations. By understanding and applying the methods and tools discussed in this module, you can provide actionable recommendations that drive business success. In the next module, we will explore how to identify areas for improvement using various analysis techniques.

Prescriptive Analysis