Price management is no small task for B2B pricing owners. While machine learning can set smarter pricing strategies, understanding how it works in the context of B2B and knowing how to deploy highly-relevant price guidance, and ultimately, create a more strategic pricing approach isn’t as straightforward. In this four-part series, Zilliant Director of R&D, Pricing Science, Amir Meimand will share a summary of how predictive analytics, prescriptive analytics and machine learning contribute to a smart pricing strategy for B2B companies.
Amir will be speaking at the 6th Annual Global Big Data Conference on August 30; he will present on the topic “An Efficient ML-Based Algorithm for Mining Substitute Products for eCommerce.”
In my last post, we discussed problems addressed by machine learning (ML) and B2B applications that are well-suited to use cases for ML. Today, I'd like to give a primer on supervised and unsupervised learning algorithms before we dive into practical applications of these algorithms in B2B.
Supervised vs. Unsupervised Algorithms
Discovering hidden patterns and structures from data often leads to actionable insights that enable various trends to be predicted and helps businesses gain a competitive advantage. Unsupervised learning algorithms are the best tools for this purpose. In general terms, these algorithms seek to find the underlying structure or pattern of a data set without knowing the answer in advance. Unlike supervised learning algorithms, unsupervised learning algorithms do not learn from an existing structure, so they are left to their own devices to discover the structure in the data.
Unsupervised learning problems can be grouped into two categories: Clustering problems and association learning.
Clustering problems are used to discover inherent groupings in the data. In other words, clustering algorithms attempt to classify data points into groups based on similar properties. For example, grouping customers based on their purchasing behavior. Several clustering algorithms in the literature can be employed for different purposes. These algorithms differ in terms of how they specify the number of clusters. For example, the k-mean, agglomerative and hierarchical clustering algorithms require the user to set the number of clusters, which requires the user to already have a good estimate of the number of clusters necessary. Others, such as affinity propagation and density-based clustering algorithms identify the number of clusters without user input.
Association learning problems are used to discover relationships between events in large dataset. An association rule extracts strong relationships; for example: “Customers that purchased X also tend to purchase Y.”
The major challenge for unsupervised learning models is evaluating its effectiveness. When evaluating them from a strictly theoretical perspective, it seems impossible for an unsupervised learning model to calculate accuracy, as there is no actual value that can be used as a comparison point. Nevertheless, a couple of approaches can be useful when evaluating the if a model is effective:
- Robustness of solution: Generally speaking, a good model should remain stable despite slight perturbations or the exclusion or addition of some data points. The volatility of the model can be measured over different scenarios.
- Benchmark: The model’s result can be compared to actual data from similar industries and areas. Zilliant customers benefit from over a decade of B2B data science expertize, and as a result, they leverage highly accurate and reliable models which are highly specialized for their own business.
In the next post, I will present a few significant applications for unsupervised learning algorithms in B2B that illustrate how unsupervised ML can help businesses grow smarter and faster.
Read Amir’s previous posts on smart pricing:
About the AuthorFollow on Linkedin Visit Website More Content by Amir Meimand