Prescriptive Analytics

Table of Contents

Prescriptive analytics is a form of advanced analytics that examines data or content to answer the question “What should be done?” or “What can we do to make __ happen?”, and is characterized by techniques such as graph analysis, simulation, complex event processing, neural networks, recommendation engines, heuristics, and machine learning.

It goes beyond descriptive and predictive analytics by not only providing insights into what is likely to happen in the future but also suggesting the most appropriate actions to take based on those predictions.

Prescriptive Analytics Methodology

Prescriptive analytics methodology involves several steps, encompassing data collection, data processing, model creation, and decision-making. Here’s a typical process:

Data Collection: The first step in any data analytics process involves gathering the necessary data. This can include historical data, real-time data, or a mix of both. The data can come from multiple different sources, such as databases, data warehouses, or external data sources.
Data Processing: After the data is collected, it’s processed and cleaned. This includes handling missing or incorrect data, normalizing data, and conducting other pre-processing steps to make the data suitable for analysis.
Feature Selection & Engineering: This step involves identifying the key variables, or features, in the data that will be used for model creation. Sometimes, existing variables might be combined or transformed to create new, more predictive features, in a process known as feature engineering.
Model Creation: The next step is to create a model that can use the selected features to predict outcomes. This could involve various statistical, machine learning, or AI techniques. Often, many different models are created and tested in this step to see which one produces the best results.
Model Testing & Validation: After the model is created, it’s tested against a validation dataset to assess its performance. Depending on the results, it may be necessary to go back to the model creation step to refine the model further.
Optimization & Simulation: Once the model is performing well, it can be used for optimization and simulation. This can involve using the model to simulate different scenarios and find the optimal solution. For instance, a company might use a prescriptive model to find the optimal price for a product that maximizes profit while keeping customer satisfaction high.
Decision Making: Finally, the results of the prescriptive analysis are used to make decisions. The goal of prescriptive analytics is to guide action, so the final step involves actually using the insights gained to make a decision or take an action.
Review & Monitoring: Post-decision, the results should be monitored to ensure they are as expected. This feedback can be used to further refine the model and improve future decision-making.

Prescriptive Analytics Techniques

Prescriptive analytics uses various techniques to help organizations decide on a course of action based on the results of descriptive and predictive analytics. Here are a few commonly used techniques:

Optimization

This is a mathematical approach used to find the best solution among various alternatives under the given constraints. Techniques like linear programming, integer programming, and more advanced multivariate algorithmic methods can be used. These methods help find the optimal solution for complex decision problems with multiple variables and constraints.

Simulation

Simulation techniques allow businesses to model a situation and then simulate different actions to see how each would affect the outcome. This can be particularly useful in scenarios where it’s difficult or costly to test different actions in the real world.

Heuristics

Heuristic methods are rule-based techniques used to speed up the process of finding a satisfactory solution, where an optimal solution can be difficult to identify. These are often used in complex or large-scale decision problems.

Machine Learning

Machine learning algorithms can be used to learn from historical data and make predictions about future data. In prescriptive analytics, these predictions can then be used to inform decisions. Machine learning can incorporate a wide variety of techniques, including regression, classification, clustering, and others.

Artificial Intelligence (AI)

More advanced prescriptive analytics solutions may use AI techniques like neural networks, deep learning, or reinforcement learning to model complex relationships and make predictions.

Decision Tree

Decision trees are schematic, tree-shaped diagrams used to determine a course of action or show a statistical probability. They present possible options, outcomes, and resource costs in a detailed, easy-to-follow format.

Monte Carlo Simulation

This technique uses probability distributions to model risk or uncertainty. It allows the analyst to run multiple ‘what if’ scenarios and understand the likelihood and impact of different outcomes.

Genetic Algorithms

These are heuristic search algorithms used for optimization and search. They are based on the process of natural selection, with concepts of mutation, crossover, and selection, and can be used to identify optimal or near-optimal solutions to complex problems.

Prescriptive Analytics Tools

There are several tools available in the market that can help with implementing prescriptive analytics. Here are a few notable ones:

IBM Decision Optimization: This is part of IBM’s prescriptive analytics portfolio. It includes CPLEX Optimizer for mathematical programming and CP Optimizer for constraint programming.

FICO Xpress Optimization: FICO offers a prescriptive analytics suite that includes mathematical optimization, decision modeling and solution design, rapid application deployment, and more.

SAS Optimization: This tool from SAS helps solve complex optimization problems. It includes linear, mixed integer, quadratic, nonlinear and general constraint optimization techniques.

Gurobi: Gurobi Optimizer is a state-of-the-art solver for mathematical programming. It supports a variety of programming languages and includes linear programming, quadratic programming, and mixed-integer programming methods.

Frontline Systems Solver: This tool provides optimization in Excel, with both linear and nonlinear programming options. It’s a great tool for those familiar with Excel and prefer to work in that environment.

River Logic: River Logic’s platform focuses on prescriptive analytics and offers scenario analysis, risk analysis, and constraint-based optimization.

RapidMiner: RapidMiner is a data science platform that supports all steps of the machine learning process, including modeling, validation, and deployment. It’s particularly popular for its user-friendly interface.

DataRobot: DataRobot offers an AI platform that supports data preparation, modeling, validation, and deployment. It has robust capabilities in predictive and prescriptive analytics.

Alteryx: Alteryx provides a range of tools for data preparation, blending, and analytics. It can handle large volumes of data and supports advanced modeling techniques.

AnyLogic: AnyLogic is a simulation modeling tool. It supports system dynamics, discrete event, and agent-based modeling, which can be used in a prescriptive analytics context.

Prescriptive Analytics Examples

Some Prescriptive Analytics examples in real life are as follows:

UPS: United Parcel Service (UPS) utilizes prescriptive analytics to optimize its package delivery operations. The company uses advanced algorithms that consider factors like package weight, size, destination, and delivery routes to optimize its fleet management and minimize fuel consumption. UPS also employs prescriptive analytics to determine the most efficient delivery sequences, reducing travel distances and improving overall operational efficiency.
American Airlines: American Airlines uses prescriptive analytics to optimize its flight scheduling and crew assignments. By analyzing historical flight data, weather patterns, and crew availability, the airline can make informed decisions on flight routes, departure times, and crew assignments. This helps them minimize delays, reduce costs, and improve customer satisfaction by ensuring efficient flight operations.
Starbucks: Starbucks utilizes prescriptive analytics to optimize its store locations and staffing levels. By analyzing data on customer demographics, foot traffic, sales patterns, and local market trends, Starbucks can determine the most suitable locations for new stores and optimize staffing schedules to align with customer demand. This enables the company to enhance customer experience and maximize profitability.
Amazon: Amazon utilizes prescriptive analytics in its fulfillment centers to optimize inventory management and order fulfillment. By analyzing real-time data on customer orders, inventory levels, and warehouse operations, Amazon can determine the most efficient picking routes, storage locations, and inventory replenishment strategies. This helps them reduce delivery times, minimize stockouts, and improve overall operational efficiency.
Chevron: Chevron, an energy company, employs prescriptive analytics for predictive maintenance and asset optimization. By analyzing sensor data from equipment, historical maintenance records, and operational parameters, Chevron can predict equipment failures, schedule maintenance activities proactively, and optimize the performance of their assets. This allows them to minimize downtime, reduce maintenance costs, and optimize production levels.

Prescriptive Analytics Case Study Example

Let’s consider a case study from the airline industry where prescriptive analytics can be effectively applied:

Case Study: Airline Revenue Optimization with Prescriptive Analytics

An international airline was struggling with optimizing its revenue due to fluctuating demand, changing prices, competition, and other factors. They turned to prescriptive analytics to help address these challenges.

Step 1: Data Collection

The first step involved gathering data. The airline collected data on past flights, including details about each flight (time, duration, route, etc.), the number of passengers, ticket prices, and more. They also gathered data about external factors such as holidays, economic indicators, and competitors’ ticket prices.

Step 2: Data Processing and Feature Selection

The data was then cleaned and processed. Variables that were likely to have an impact on the ticket sales and revenue (like time of flight, price, route popularity, competition prices, etc.) were selected for further analysis.

Step 3: Model Creation

The airline then created a predictive model to forecast demand for each flight based on the variables selected. After that, they developed a prescriptive model that used this demand forecast to determine the optimal ticket price. This model considered various constraints, such as the need to cover costs and the prices being offered by competitors.

Step 4: Model Testing & Validation

The model was then tested and validated using a subset of their data. The model’s results were compared against actual outcomes to assess its performance.

Step 5: Optimization & Simulation

Once the model was validated, it was used for optimization and simulation. The airline was able to simulate different pricing strategies and see how each one would affect their revenue.

Step 6: Decision Making

Based on the results of the prescriptive analysis, the airline adjusted their pricing strategy. They were able to set prices that maximized their revenue while still being competitive in the market.

Step 7: Review & Monitoring

After implementing the changes, the airline continued to monitor the outcomes to ensure they were in line with the model’s predictions. The model was periodically updated with new data to ensure it remained accurate.

As a result of implementing prescriptive analytics, the airline was able to significantly increase its revenue. It also gained valuable insights that helped it understand the key drivers of demand and revenue in its business. This case study demonstrates how prescriptive analytics can be used to guide decision making and optimize outcomes.

When to use Prescriptive Analytics

Prescriptive analytics is beneficial when you need to make decisions in complex situations, where there are many variables and possible outcomes to consider. Here are several situations when prescriptive analytics can be particularly useful:

Supply Chain Optimization: If a company wants to optimize its supply chain operations – like inventory management, distribution, and logistics – prescriptive analytics can help determine the best course of action considering multiple factors such as demand forecasts, cost, time, and resource constraints.

Pricing Strategy: When a business wants to set pricing for its products or services, prescriptive analytics can help find the optimal price point that maximizes profit while considering factors such as demand elasticity, competition, production cost, and customer willingness to pay.

Workforce Scheduling: If an organization needs to schedule its workforce effectively – like in hospitals or call centers – prescriptive analytics can help. It can take into account various factors such as employee availability, skills, labor laws, and forecasted demand to create optimal schedules.

Resource Allocation: When a business or organization has limited resources that need to be allocated across different projects, units, or initiatives, prescriptive analytics can help optimize this allocation to achieve the best possible outcomes.

Risk Management: In situations where businesses need to manage various types of risks, prescriptive analytics can be beneficial. By simulating different scenarios, it can provide insights into how to mitigate risks and what actions should be taken in different potential future situations.

Marketing Optimization: If a company is trying to optimize its marketing campaigns, prescriptive analytics can provide insights into which strategies will produce the best results. It can consider factors such as customer preferences, budget constraints, and past campaign performance.

Strategic Planning: For businesses devising long-term strategic plans, prescriptive analytics can offer insights into potential future scenarios, helping the business to create effective strategies for growth, investment, and more.

Applications of Prescriptive Analytics

Prescriptive analytics has a wide range of applications across multiple industries, helping businesses and organizations optimize their decisions and operations. Here are some common applications:

Supply Chain Management: Prescriptive analytics can help optimize various aspects of the supply chain, such as inventory levels, distribution routes, and warehouse locations. It can suggest the best course of action to maximize efficiency and minimize costs based on multiple variables like demand forecasts, transport costs, and delivery times.

Healthcare: In healthcare, prescriptive analytics can be used to optimize patient care and health outcomes. It can analyze patient data and suggest personalized treatment plans, optimal staffing schedules in hospitals, or effective ways to manage resources during health crises.

Retail: Prescriptive analytics can help retailers optimize pricing, manage inventory, personalize marketing, and even design stores. For example, by analyzing customer behavior data, it can recommend strategies for improving sales or customer experience.

Finance: Financial institutions can use prescriptive analytics for portfolio optimization, risk management, fraud detection, customer segmentation, and more. It can suggest optimal investment strategies based on market trends, regulatory constraints, and investor preferences.

Telecommunications: Prescriptive analytics can help telecom companies in network optimization, predictive maintenance, customer churn prediction, and personalized marketing. For example, it can suggest where to upgrade network infrastructure or how to target promotional offers to customers.

Transportation and Logistics: Prescriptive analytics can help optimize routes, schedules, and resources in transportation and logistics. For example, it can suggest the most efficient routes for delivery trucks considering traffic conditions, delivery deadlines, and fuel costs.

Energy and Utilities: Energy companies can use prescriptive analytics for demand forecasting, optimizing energy production, and predictive maintenance of infrastructure. For instance, it can suggest when and where to perform maintenance on power lines or how to adjust production based on demand forecasts and resource availability.

Manufacturing: Prescriptive analytics can be used for production planning, quality control, predictive maintenance, and supply chain optimization in manufacturing. For example, it can recommend optimal production schedules based on demand forecasts, raw material availability, and production costs.

Advantages of Prescriptive Analytics

Prescriptive analytics can provide numerous benefits to organizations across different industries. Here are some key advantages:

Informed Decision Making: Prescriptive analytics provides actionable recommendations based on data, helping decision-makers understand the potential impact of their choices before they make them. This leads to more informed and effective decision making.
Improved Operational Efficiency: By optimizing processes and resource allocation, prescriptive analytics can lead to increased operational efficiency. This can result in cost savings and improved productivity.
Increased Profitability: Through applications like price optimization, inventory management, and improved resource allocation, prescriptive analytics can help businesses increase their profitability.
Risk Management: Prescriptive analytics can help organizations anticipate and manage risks more effectively. By modeling different scenarios and outcomes, it can provide insights into how to mitigate risks and what actions should be taken in different potential future situations.
Personalization: In areas like marketing and customer service, prescriptive analytics can help create personalized experiences, which can lead to increased customer satisfaction and loyalty.
Competitive Advantage: Organizations that effectively leverage prescriptive analytics can gain a competitive advantage by making more effective decisions and optimizing their operations.
Future Readiness: Prescriptive analytics not only deals with the present but also anticipates future scenarios. By doing so, it prepares organizations for potential future challenges and changes, making them more resilient and adaptable.
Time-Saving: Prescriptive analytics can automate complex decision-making processes, saving valuable time and allowing human decision-makers to focus on strategic, creative, or unstructured problems.

Disadvantages of Prescriptive Analytics

While prescriptive analytics offers many benefits, it’s not without its challenges. Here are some potential disadvantages:

Data Quality: Prescriptive analytics relies heavily on the quality and completeness of data. If the data used is inaccurate, incomplete, or biased, the recommendations made by the prescriptive model can be flawed, potentially leading to poor decisions.
Complexity: The models used in prescriptive analytics can be highly complex, often requiring expertise in advanced statistics, machine learning, and business strategy. Not all organizations may have access to such expertise.
Cost: Implementing prescriptive analytics can be costly. This includes costs for data collection and storage, analytics software and tools, and hiring or training staff with the necessary skills. For some organizations, particularly smaller ones, these costs may be prohibitive.
Time-consuming: The process of setting up a prescriptive analytics system can be time-consuming. It involves gathering and cleaning data, building and validating models, and integrating the system into the organization’s decision-making process.
Data Privacy and Security: Prescriptive analytics often involves handling large volumes of sensitive data, which raises concerns about data privacy and security. Organizations must ensure they comply with all relevant regulations and best practices to protect their data.
Resistance to Adoption: Like any change, the introduction of prescriptive analytics can face resistance from staff, particularly if they feel their jobs are threatened or they don’t understand the benefits of the new approach.
Over-reliance on Analytics: While prescriptive analytics can be powerful, it’s essential to remember that it should support, not replace, human decision-making. There’s always a risk of becoming over-reliant on analytics and neglecting the importance of human intuition, judgment, and understanding of context.