Machine learning, data science, artificial intelligence – by now it is safe to say that these terms are more than just short-term trending topics. Tools based on machine learning (ML) technologies are here to stay and they might even revolutionize whole business sectors
The progress in hardware technology resulting in an increased available computing power have enabled the breakthrough in available computing power has allowed for progress in machine learning and related areas. As a result, research and development interests have been strongly focused on these areas and advancements have been fast and numerous in the past 10 to 15 years.
But what is machine learning all about? The aim of machine learning is to let machines act without being explicitly programmed. Machines are rather presented with data regarding a certain topic or a problem at hand and are asked to draw their own conclusions on the basis of the data. Thus, they are implementing workflows and actions themselves.
While there is a tendency to use terms such as artificial intelligence or data science synonymously with machine learning, this is not completely true. In contrast to machine learning, artificial intelligence is the science and engineering of making intelligent machines. Machine learning is one tool that facilitates artificial intelligence. Whereas, data science (in the past mostly called business analytics or business intelligence) is aimed at extracting knowledge and insights from data by using scientific methods, processes and algorithms. Machine learning approaches are again a tool of data science and are used to detect and understand patterns in the data.
Machine Learning and Predictive Analytic
In predictive analytics facts represented by current and / or historical data are analyzed to derive a prediction about the future. In order to analyze data statistical techniques including machine learning approaches are used.
In the past 20 years, predictive analyses have already been widely used, mainly within applications of the finance sector. An example for such an application is the well-known credit scoring which tries to predict whether a potential customer will make future credit payments on time and ranks potential customer on this estimated likelihood.
Generally speaking, predictive analytics can be used in business to identify risks and opportunities. Analytical customer relationship management is just one popular business application. The use of Predictive Machine Learning in this regard allows a company to focus their efforts in marketing and sales efficiently by identifying promising prospects. Moreover, predictive analysis can, for example, be used in the planning of a marketing strategy by identifying the most effective combination of product versions, marketing material, communication channels and timing.
Small but Powerful
But which data do you use in a predictive analysis? Does every business produce data that can be used that way?
In the past ten years predictive analysis often went hand in hand with big data. For the last decade we saw big data as the ultimate chance to gain insights and detect patterns. It has been applied to predict customer behavior, market trends or even election results. Nevertheless, big data requires quite some effort – “nomen est omen”. Not only do we need huge amounts of data, the analysis of it requires experts such as data scientists.
Small data on the other hand is data that is considered “small enough for human comprehension”. This is the kind of data that almost every business holds. mall data is everywhere and can be collected and analyzed by individual or departments.. The good news: For many problems and questions, small data in itself is enough.
While small data is comprehensible for us as humans, you might still want to use some tool or application to analyze it. However, compared to big data, we can work with small data in a more intuitive way using tools and apps instead of depending on the assistance of data science experts.
A Farewell to Excel?
“Everything processed in Excel is small data.”
But this does not mean you need to use Excel to process small data – this is only one way to go. If you are already used to working with data but you do not have the technical skills to build predictive models, you might still rely on spreadsheets to get the job done.
Some go as far as to consider Excel a powerful data science tool. With Excel being around since the late 80s it definitely is a data science pioneer. But when it comes to scripting languages and predictions it feels outdated. Simply put, the predictive power of a spreadsheet is limited. For a prediction approach you might want to part from Excel and consider a machine-learning-based solution instead – ideally one that does not require you to develop your machine learning skills first.
One such technology that works very well with small datasets is MyDataModels. It does not require data science or machine learning skills – not even coding skills are necessary to get started with MyDataModels since it has been created with the goal to democratize machine learning. What is more, you do not need a powerful server – your laptop, or even your mobile will do.
What do you need to do to analyze your (small) data using MyDataModels? Just follow these six steps:
- Prepare your dataset and upload it for the training.
- Define your business objective, the target variable: What do you want to predict?
- Rank your explanatory variables by predictive power, from highest to lowest to assess quality of your dataset.
- Start the model creation with the click of a button. But give it time – depending on the size of your dataset it may take some hours to create your model.
- Assess model’s performance against your business objectives.
- Run your model on a new dataset and start predicting, or restart process at step 1 with additional variables – to increase data quality – if model’s performance is not satisfactory.
Who knew that building Predictive Machine Learning can be done without coding skills?