9.8 Special Topics in Forecasting
As we journey through the intricacies of forecasting, we’ve explored a plethora of methods and collaborative techniques. However, to truly excel as forecasters, there are two pivotal areas that warrant our attention: understanding modern forecasting methods and mastering the art of forecasting new products.
9.8.1 Modern Forecasting Methods
In the dynamic intersection of business and technology, forecasting methods have seen remarkable advancements. Traditional methods, while still prevalent and valuable, are being complemented by cutting-edge techniques rooted in neural networks, machine learning (ML), and artificial intelligence (AI). These modern methods harness vast data and sophisticated algorithms to offer potentially more accurate predictions.
Neural Networks: Neural networks are computing systems inspired by the human brain’s structure, consisting of layers of interconnected nodes (or “neurons”) that process information. They excel at recognizing patterns in large datasets, making them invaluable for complex forecasting tasks.
Example: Consider a global fashion retailer trying to predict the demand for a new line of clothing. A neural network could analyze not just previous sales data, but also patterns in online search trends, social media mentions of related fashion terms, and even images from fashion shows. By training on this diverse set of data, the neural network can identify nuanced patterns and trends that might be missed by traditional methods.
Machine Learning (ML): ML, a subset of AI, enables computers to learn from data. Instead of explicit programming, an ML model trains itself using data to make predictions. In forecasting, ML can sift through historical data, pinpoint patterns, and refine its predictions as more data is fed into it.
Example: An e-commerce platform might use ML to forecast sales of a particular product. The model could consider past sales, user reviews, website traffic, and even the time users spend on a particular product page. Over time, as it gathers more data points like seasonal shopping trends or promotional events, the ML model refines its predictions.
Artificial Intelligence (AI): AI, a broader concept encompassing ML, aims to create machines mimicking human intelligence. In forecasting, AI can amalgamate various data sources, analyze them, and even adapt its forecasting strategy based on changing variables.
Example: A supermarket chain could use AI to forecast the demand for perishable goods. The AI system might factor in past sales data, weather forecasts (as people might buy more ice cream on hot days), local events, and even data from news sources (like a health scare related to a product).
The primary advantage of these modern methods is their capacity to process vast data and intricate interrelationships, which might be overwhelming for traditional methods. However, they also demand more sophisticated infrastructure and expertise. And, crucially, their predictions hinge on the quality of the data they’re trained on. Overall, the realms of neural networks, ML, and AI might seem intricate, but their potential in forecasting is vast. As technology continues its rapid evolution, these methods will likely become indispensable in the forecaster’s toolkit, driving more nuanced and informed business decisions.
9.8.2 New Product Forecasting
A common assumption in forecasting is that new products, by virtue of being “new”, lack historical data, pushing us towards qualitative forecasting methods. However, this isn’t always the case. New products can be categorized into four distinct types:
- Extension: Refers to a new product or service based on an existing one. An example would be a tech company releasing an upgraded version of its flagship smartphone.
- New Geography: Pertains to a product or service launched in a new market. A food company introducing a product line tailored for the Chinese market exemplifies this.
- New to the Company: Denotes a product or service that a company introduces for the first time. A software firm venturing into cloud computing services would fall under this category.
- New to the World: Represents truly innovative products or services that are unprecedented. The debut of the first smartphone is a classic example of a new-to-the-world product.
It’s crucial to note that it’s primarily the “new to the world” products that genuinely lack historical data. For the other categories, while the product might be “new” in some sense, there’s often related data that can be leveraged for forecasting.
