Data science has always been our core business (we even started in 2006, long before the hype). The large majority of our work evolves around prediction in a business context. For example, we predict which prospects to attract, what customers will buy next, which customers will leave, who will pay, which machine will fail, who risks a burnout, etc.
Yet besides prediction, we also gathered a lot of experience in creating customer segmentations and recommendation engines. We nailed projects in process mining and demand forecasting, where we optimize process efficiencies and logistics.
And while we are often included to create the first success story, we have great references in industrialising analytics in mature organisations.
With the growing popularity of data science, the scope of the domain has broadened. Where our original projects were often well-defined and well-structured, we see an increasing demand for data exploration projects. In a growing number of projects, we work in an agile way on exploring new data in a variety of formats and sources, often using new technology. These new data and technologies lead to novel applications such as image recognition, IoT analytics, location-based analytics, network analytics, etc.
We ensure we continuously innovate through dedicated development time, our connection with modern data science communities and our collaborations with academic programs and business schools.
Since the very start, we invested a lot of time in training and coaching for both managers and data scientists where we focus on the essential components of data science projects. We consider training to be an important factor for increasing adoption and maturity of data science.
We often conduct analytical roadmaps, leading to a concrete list of feasible and valuable project ideas. We offer intuitive, non-technical training for managers, as well as technical hands-on training for data scientists. And we coach both managers and data scientists in their challenges.
Subscribe to our training newsletter to receive instant notifications of newly scheduled open trainings, or contact us at firstname.lastname@example.org to receive a quote for in-house training for managers or data scientists.
A fantastic customer experience is of strategic importance for many companies today. In this age of information overload, customers expect to be targeted adequately, while marketing efforts are increasingly evaluated based on their return on investment.
Through Data Science projects, marketeers today are increasing targeting efficiency and relevance for different types of marketing campaigns over different channels (online, offline) and media. Typical applications in the marketing domain focus on new customer acquisition, cross-selling, upselling, customer retention and (re)activation.
In risk evaluations, it is widely understood that mere human judgment does not suffice to adequately estimate future risk.
Typical applications in the risk domain include credit risk (i.e. estimation of creditworthiness) and fraud risk (i.e. detection of fraudulent behavior). A growing range of organizations today make use of proven data-driven methodologies to analyze massive amounts of data, in order to provide reliable estimations of future risk.
Also in operations, a growing number of organizations deploy the abundance of internal data with modern data science and machine learning techniques in order to improve process efficiency. Such applications focus on optimizing stock volumes, maximizing production output or minimizing inputs, waste, pollution, etc.
Data Science has the leveraging potential to turn the currently existing massive information flows into value.
The newest applications in advanced analytics lie in the domain of human resources. Indeed, an increasing number of companies is deploying advanced and predictive analytics to support decision making in HR.
The most obvious applications are related to planning and forecasting, recruitment, development and retention of staff. We are extremely excited about our first experiences in this field.
01 Data science or predictive analytics?
Many successful data science projects contain an important predictive component. Predictive analytics is a term more often used for challenges in a business context.
It refers to the methodology needed to convert historical information into valuable and accurate predictions of future events, in order to improve decision making. In this way, banks aim to predict who will repay credits, retailers aim to predict who will visit the store, and which item they will buy next. Telecom operators predict the probability of a client leaving the company, etc.
02 Do you only use Python?
No, the name Python Predictions originates from the Greek mythology, where the Greek god Python resided at the Oracle at Delphi. Hence, it was the first Greek god capable of predicting the future.
Python, as a coding language, has become increasingly popular recently, also in our team. But we equally use R, or even commercial software packages such as SAS or IBM SPSS. In short, we use the solution that makes most sense for a specific client.
03 How can we assess the opportunities?
Most clients address us with a specific business problem. For example, some clients aim to improve customer retention, others aim to improve new client acquisition, others aim to reduce fraudulent behavior and others aim to optimize stock levels. However, Python Predictions has also gained experience with more general questions, such as: is my company mature to benefit from predictive analytics. The time needed to perform such assessments largely depends of the strategic importance of analytics, and may require as little as one meeting or as much as a week.
04 How much internal expertise is required?
While some of our clients have a large team of internal analytical experts, we have equally worked with companies that have no previous internal experience in terms of predictive analytics. In the different settings, we work towards the same goal: offering powerful predictions that can be interpreted and trusted by the business users. This means that we always strive for an optimal solution, but we aim to adapt our communication and presentation to the interests of the user, balancing the focus between commercial and/or technical aspects.
05 How can we collaborate practically?
Depending on the availability of hardware, software and office space, we can perform our projects at our client’s location or at our offices. Our primary goal is to perform well-defined analytical projects from A to Z, i.e. starting from the definition of the business problem to the solution. Furthermore, we also offer support and guidance during the implementation of the analytical solution in the customer’s IT infrastructure. An initial data assessment may be useful to identify and quantify opportunities.
We also offer company-specific training and coaching in the areas of predictive analytics and segmentation. These courses are adapted towards the (desired) level of expertise and interests of the client.
06 Is predicting behavior possible?
Predicting the future is challenging. It is like predicting tomorrow’s weather. Knowing that there is an 80% chance of sun does not necessarily mean you will stay dry without an umbrella. The great advantage of predictive analytics however, lies in the fact that one does not make only one prediction. In marketing, for example, predictions of the behavior of every single customer can be made for a large customer base, whereby the joint performance will be related to the return on investment. Hence, when predicting behavior, it should not be expected that all predictions turn out 100% correct, but in important projects, a small difference in predictive quality can be reflected in large monetary benefits.
07 Do you invest in analytical communities?
Absolutely. Since the launch of Python Predicitons, we have invested both time and resources in analytical communities. From 2006 to 2009, we were heavily involved in building baqmar.eu, a large vendor-neutral Belgian community for analytics practitioners. More recently, from 2012 to 2015, Python Predictions was responsible for programming Predictive Analytics World London, an international vendor-neutral event focused on bringing together managers and practitioners in the domain of Predictive Analytics. Since 2015, we are actively collaborating with the Brussels Data Science Community where we offer executive training.
08 Is predictive analytics useful for my industry?
Predicting customer behavior is useful across a wide variety of industries and applications. Although retailing, e-tailing, mail-order, banking, telecommunications and subscription services provide some of the most obvious players that currently profit from predictive modeling, the techniques are generic and industry-independent. However, we consider an accurate definition of the business question and an understanding of a company’s (marketing) strategy and tactics crucial to deliver valuable results.
09 What are the necessary requirements?
Some requirements exist to allow for predicting future events. In order to predict the future, one must gain a good understanding of the past. This will very often imply that it is necessary to recreate the full history of similar events in the past. It is only by understanding previous behavior that one can make informed decisions about future behavior. Contrary to popular belief, the methodology of predictive analytics does not apply to Big Data problems only. There is no absolute minimum of records needed to engage in valuable predictive analytics projects.