Predict or Decide? Mastering the Art of “Will It Happen” vs “Should It Happen”

In everyday life we encounter two common types of questions: "will it happen?" – a prediction problem, and "should it happen?" – a decision problem. Although related, their essence, solution methods, and required tools differ significantly. This article explores how to distinguish these questions and proposes methodologies to address them effectively.

1. Clarify Problem Nature: Prediction vs Decision

When solving any problem, first determine whether it belongs to the predictive "will it happen" category or the normative "should it happen" category.

1. “Will it happen?”: Future Uncertainty

This is essentially a forecasting problem that requires using existing data, models, or trends to make reasonable judgments about future possibilities, such as disease outbreaks or market share growth. It typically involves statistics, mathematical modeling, and machine learning to reduce prediction error and improve accuracy.

2. “Should it happen?”: Value Judgment of Action

This is a decision problem that demands choosing among alternative actions by weighing benefits, ethics, legality, and social values, such as investing in a new industry or adopting a public policy. It relies more on value judgments, ethical analysis, and risk‑assessment tools.

2. Methodology for Solving “Will it happen?” Problems

The core of a "will it happen" problem is to make the most accurate forecast possible based on available information.

1. Data Collection and Processing

Gather relevant data, clean and preprocess it to ensure quality.

2. Model Selection and Tuning

Choose appropriate models based on data characteristics and problem context—for example, time‑series models for continuous temporal data or classification models for discrete events. Model choice should be guided by the nature of the data and the problem.

3. Result Validation and Adjustment

Evaluate predictive performance; if results are unsatisfactory, adjust parameters or select a more suitable model.

Case Study: Predicting a City’s Future Air‑Pollution Levels

Assume a city wants to forecast air‑pollution trends for the next five years to plan mitigation measures. The workflow includes:

Data Collection : Gather ten years of pollution data (PM2.5, PM10), meteorological data, and industrial emission records.

Model Selection : Use a time‑series model such as ARIMA combined with an LSTM neural network to capture long‑term trends and short‑term fluctuations.

Result Validation : Compare model predictions with historical data to ensure accuracy, then produce a five‑year pollution trend for decision‑making.

3. Methodology for Solving “Should it happen?” Problems

The core of a "should it happen" problem is to weigh multiple possible outcomes and select the optimal action.

1. Define Goals and Constraints

Clarify objectives (e.g., maximize economic benefit, minimize social impact) and constraints (budget limits, legal requirements).

2. Design Action Plans

Develop multiple feasible action alternatives as a basis for analysis.

3. Multi‑dimensional Evaluation

Apply methods such as Cost‑Benefit Analysis (CBA) or Multi‑Criteria Decision‑Making (MCDM) to assess each alternative across economic, social, and environmental dimensions.

4. Value Judgment and Selection

Based on analysis, incorporate value judgments and stakeholder input to choose the optimal solution.

Case Study: Implementing a New Congestion‑Pricing Policy

A city plans a congestion‑pricing scheme but faces public controversy. The steps are:

Goals & Constraints : Reduce traffic congestion and improve city efficiency while staying socially acceptable.

Design Plans : Create various pricing levels and scopes, such as peak‑hour charges or zone‑based fees.

Evaluation & Analysis : Use simulation models to predict traffic flow impacts and assess effects on different income groups.

Value Judgment : Combine data and social surveys to select a scheme that balances efficiency and fairness.

4. Distinguish and Integrate: Using Prediction to Inform Decision

Although "will it happen" and "should it happen" are distinct, they often complement each other in practice. Forecast results provide a basis for decisions, and the outcomes of decisions can, in turn, validate the accuracy of predictions.

For example, climate scientists predict future greenhouse‑gas emissions ("will it happen"), and governments decide whether to adopt stricter mitigation policies ("should it happen"). Prediction accuracy directly influences decision effectiveness, while decisions must consider ethical, economic, and social factors.

Distinguishing between these two problem types is crucial for tackling complex issues. The former focuses on future uncertainty; the latter reflects value judgments and action choices. By applying appropriate prediction and decision‑making methodologies, we can understand problems more scientifically and choose actions more rationally. This approach is applicable not only to scientific research and policy making but also to corporate management and personal development.

In my book "Modeling: The Mathematics of Thought" I introduce four problem types—Description & Understanding, Estimation & Prediction, Evaluation & Decision, and Causation & Explanation—collectively called the DEED framework, each with its own modeling solutions.