Forecasting process
● A process is a series of connected activities that transform one or more inputs into one
or more outputs.
● All work activities are performed in processes, and forecasting is no exception.
● The activities in the forecasting process are:
○ 1. Problem definition
○ 2. Data collection
○ 3. Data analysis
○ 4. Model selection and fitting
○ 5. Model validation
○ 6. Forecasting model deployment
○ 7. Monitoring forecasting model performance
● These activities are shown in Figure 1.12.
1. Problem Definition
● Clearly define the objective of the forecast.
● Identify what needs to be forecasted (e.g., sales, demand, weather patterns).
● Determine the scope, time horizon (short-term, medium-term, or long-term), and the
accuracy level required.
● Understand the key factors influencing the problem, constraints, and stakeholders'
requirements.
● Questions that must be addressed during this phase include
○ The desired form of the forecast.
○ The forecast horizon or lead time,
● What level of forecast accuracy is required in order to make good business decisions.
● Level of aggregation of the forecast (population level / individual level) must also be
considered.
2. Data Collection
● Gather relevant historical data from reliable sources (e.g., company databases, market
research, external reports).
● Not all historical data are useful for the current problem so information collection and
storage methods and systems change over time.
● Missing values of some variables, potential outliers, or other data-related problems must
be dealt properly.
● Data types include quantitative (numerical) and qualitative (opinions or expert
judgments).
● Ensure data quality by addressing issues like missing values, inconsistencies, and
errors.
3. Data Analysis
● An important preliminary step.
● Explore and analyze the collected data to identify patterns, trends, and seasonality.
● Use statistical tools (e.g., mean, variance, correlation) and visualizations like graphs and
charts to understand the data.
● Clean the data by handling outliers, missing values, and noise to ensure it is suitable for
forecasting.
● Recognizable patterns, such as trends and seasonal or other cyclical components
visually inspected from constructed time series plots.
● A trend evolutionary movement, either upward or downward, in the value of the variable
long-term or more dynamic and of relatively short duration.
● Seasonality the component of time series behavior that repeats on a regular basis, such
as each year.
● Data smoothing performed; Numerical summaries computed; potential outliers should be
identified and flagged.
4. Model Selection and Fitting
● Choose the appropriate forecasting technique based on the data and problem. Common
methods include:
○ Qualitative methods (e.g., Delphi method, market surveys).
○ Quantitative methods (e.g., time series analysis, regression models, ARIMA,
machine learning models).
● Fit the chosen model to the data by training it to recognize patterns and relationships.
● Choosing one or more forecasting models and fitting the model to the data.
● Fitting estimating the unknown model parameters, usually by the method of least
squares.
● Several types of time series models available appropriate model chosen.
● Eg. AR, MA, ARMA, ARIMA, SARIMA, VAR, NN, Spectral Analysis Models, Bayesian
Time Series Models
● Assumptions made (hypotheses).
5. Model Validation
● Test the model using a validation dataset to evaluate its accuracy.
● Use performance metrics such as:
○ Mean Absolute Error (MAE): Average magnitude of errors.
○ Root Mean Squared Error (RMSE): Penalizes large errors more.
○ Mean Absolute Percentage Error (MAPE): Error as a percentage.
● Adjust the model if necessary to improve its predictive performance.
● widely used method data splitting data are divided into two segments—a
○ fitting segment and a forecasting segment
● The model is fit to only the fitting data segment, and then forecasts from that model are
simulated for the observations in the forecasting segment.
6. Forecasting Model Deployment
● Implement the validated model in a real-world environment.
● Ensure that the model integrates with existing systems and workflows.
● Use the model to generate actionable forecasts for decision-making (e.g., sales
forecasts for inventory planning).
● Getting the model and the resulting forecasts in use by the customer.
● Model maintenance making sure that data sources and other required information will
continue to be available to the customer.
7. Monitoring Forecast Model Performance
● Continuously monitor the model’s performance over time.
● Update the model with new data to maintain accuracy and relevance.
● Identify and address any changes in data patterns or external conditions that may affect
the forecast.
● Conduct regular performance reviews to ensure the model meets the desired objectives.
● Model may deteriorate in performance will result in larger or more systematic forecast
errors.
● Hence monitoring of forecast errors is an essential part of good forecasting system
design
