Upload a spreadsheet with a table of data, select your KPI and generate a DataStory with actionable results without being a Data Science expert
The DataStories Matlab® Toolbox allows you to load a spreadsheet with a table of data that can contain hundreds of columns with variables that can be numeric as well as text. This data is automatically analysed for type and missing data content. Numeric variables are scaled automatically, categorical variables are encoded as numeric variables and added to the data-set. The entire data-set is then screened to determined the 'health' of all available data vectors which can be analysed in detail in a data health explorer that can help you get an overview of hundreds of variables and detect problems quickly.
After selecting your KPI, the variable you care about most and would like to optimize, you are ready to determine relevant relationships and build your predictive models. Linear and non-linear relationships between variables are quantified and represented in an intuitive, interactive explorer. Maximally predictive models can be built with a push of a button without worrying about over-training or what specific settings to use. An appropriate model-complexity is determined for your specific data-set and feature selection is happening behind the scenes such that only the set of necessary and sufficient number of variables become part of your model. These models can be inspected in detail and interactive what-if explorers can be generated that allow you to quickly explore the interactions and effects of your remaining variables. All models are actually ensembles and will produce a prediction as well as a local confidence limit that varies depending on how much variation there is in the underlying models.
Detailed reports can be generated that document all aspects of your modelling process. Your modelling results can be organised in projects, DataStories and iterations within a DataStory. The DataStories Matlab® Toolbox can also be used in combination with the DataStories Cloud platform that does all computations remotely and has a very easy to use interface in your favorite browser.