Deep dive into Qlik Cloud and learn more about Machine Learning
- by Michiel Hofsteenge
"Qlik Cloud saved my organization!" Of course, that is one of the best compliments you can get as an organization from a customer. Why this customer is so enthusiastic, I like to explain in my 7-part blog about the possibilities and content of Qlik Cloud. My name is Michiel Hofsteenge, Tech Lead Data Analytics at Cmotions and Qlik Cloud fanatic.
Machine learning as an extension of your analytics
Recently, we have been reading more and more about Artificial Intelligence(AI) and Machine Learning (ML). For some still a vague concept. Yet the impact of these techniques will become noticeable for everyone very soon. The world of business intelligence is changing along with it just as fast. Where deep knowledge used to be required to be able to do anything and everything, it is increasingly at your fingertips. We are writing anno 2023. The year of the AI revolution.
Of course, it started a few years ago with the insight advisor. AutoML is one of the latest techniques added to that. Since it's still something completely different from what you're used to with the Qlik analytics package, it's definitely worth taking a closer look at.
Machine learning always sounds enormously complicated, but especially with today's technologies, it has been made enormously easy for an end user in terms of technology. So is Qlik's AutoML. What really matters is: do you understand what you are doing?
Reload tasks
Let me start by explaining the basics of Machine Learning. ML is a part of Artificial Intelligence. Actually a bit of the coathanger under which everything around smart machines hangs. You can think of it as making machines (read: computer) just as smart as people. The advantage of computers is that they can perform actions many times faster than people. Pure profit!
Most people would like to have a glimpse into the future. Having certainty is one of the basic needs of a human being. For you, analyzing the future is most likely very useful. Approximately what will be your sales next year? How much effect is a marketing campaign going to have? Numerous questions that can be answered by looking at the future with data. Machine learning takes this out of your hands.
Everything is based on algorithms. Basically, you don't need to know in detail what each algorithm calculates. It is certainly useful to learn a little about how algorithms work before you start Machine Learning in Qlik.
By using the right variables, a prediction can be made. Based on this prediction, you can act again to make better choices. Determining the right variables is often quite a task. And once you have determined these variables, do you have the right data at hand? Fortunately, external data is often available in addition to internal data. One example is data that can be obtained free of charge from Statistics Netherlands (CBS).
Automated Machine Learning
Automated Machine Learning, or AutoML, is an approach to machine learning that automates the process of building and implementing predictive models. It allows users to use complex algorithms and techniques without in-depth technical knowledge. With AutoML, organizations can take advantage of the power of machine learning even if they do not have extensive experience with data science.
AutoML uses an intuitive and user-friendly interface to reduce the complexity of building predictive models. It follows a simple, step-by-step process that even users without advanced technical skills can follow. Let's take a look at the key steps in the process:
1. Data exploration
With AutoML, you can easily upload and explore data. The platform provides tools to visualize the data, understand the distribution of variables and identify potential relationships between variables.
2. Preprocessing of data
Before you can build models, you may need to perform some preprocessing operations on the data. For example, dealing with missing values, normalizing data and converting categorical variables. Qlik's AutoML provides built-in data preprocessing functionality to facilitate this process.
3. Model selection
Based on your objectives and the type of problem you want to solve (classification, regression, clustering, etc.), AutoML automatically selects the most appropriate models to explore. It uses advanced algorithms to identify the best-performing models and provides a list of recommendations.
4. Model training and evaluation.
Once you have selected the models you want to explore, AutoML performs training and evaluation. It divides your data into training and test sets, applies the selected models and evaluates their performance against relevant metrics, such as accuracy, precision and recall.
5. Deployment and implementation
Once you are satisfied with the performance level of a model, it can be easily implemented within your environment. AutoML from Qlik supports seamless integration with other Qlik products and provides opportunities to share predictions and insights through dashboards, reports and other analytics applications.
In conclusion
Ultimately, AutoML offers considerable capabilities. Unfortunately, it is not (yet) as comprehensive as, say, Azure. But it certainly provides a nice play environment to take the first steps toward a more mature BI environment. Is it an Analytics plus solution, yes perhaps. But certainly not wrong to have with it.
