ML-Blood Sugar Level Example

You can find the source for the example here:

Here we used CLS-Luigi to create a ML Pipeline to predict the blood sugar level of some patients. The Steps are very easy to understand. We start by loading the dataset from Scikit-Learn, then we split it into 2 subsets for training and testing.

The first variation point is the scaling method. We introduce 2 concrete implementation, namely RobustScaler & MinMaxScaler. After scaling we have our second variation point which is the regression model. Here we have also 2 concrete implementation, namely LinearRegression & LassoLars.

Lastly we evaluate each regression model by predicting the testing target and calculating the root mean squared error.

variant_1.py and variant_2.py both implement the same pipeline, but in variant_2.py we tested out to use dictionaries for the required and output methods of tasks to get overall a more understandable and easier-to-read source code. For more information on that, check out the getting started tutorial

Requirements

The example contains a requirements.txt file. To experiment with the example, you can set up your environment by executing the following command:

# cd into the ml blood sugar level example folder
pip install -r requirements.txt

Static Visualization

Dynamic Visualization