FEDOT Framework quick start guide

How to install

pip install fedot

How to create your own composite model in manual way

  • Step 1. Specify problem type and choose dataset.

import pandas as pd

model = Fedot(problem='classification')

dataset_to_train = pd.read_csv(train_file_path)
dataset_to_validate = pd.read_csv(train_file_path)
  • Step 2. Create Pipeline instance, create nodes with desired models

node_first = PrimaryNode('logit')
node_second= PrimaryNode('xgboost')
node_final = SecondaryNode('knn', nodes_from = [node_first, node_second])
pipeline = Pipeline(node_final)
  • Step 3. Fit the pipeline using fit method.

model.fit(features=dataset_to_train, target='target', predefined_model=pipeline)
  • Step 4. Obtain the prediction using predict method.

prediction = model.predict(features=dataset_to_validate)

How to compose the pipeline in automated way

auto_model = Fedot(problem='classification')
pipeline = auto_model.fit(features=dataset_to_train, target='target')
prediction = auto_model.predict(features=dataset_to_validate)
auto_metrics = auto_model.get_metrics()

How to setup the development environments for the Fedot

  • Step 1. Download FEDOT Framework.

    • First of all, you need to clone the FEDOT Framework to your personal computer. You can do it directly using the button ‘clone or download’ (red square) or you can install IDE (e.g. PyCharm) and using the ‘clone in Pycharm’ button (blue square), which will open the files you need directly in the Pycharm project.

    • For more details, take a look at the picture below.

      Step 1

  • Step 2. Creating VirtualEnv in Pycharm project.

    • Next, you need to create virtual enviroment in your Pycharm project. To do this, go through the following sections: ‘File - Settings - Project Interpreter - Add new’.

    • For more details, take a look at the picture below.

      Step 2

    • After you have created a virtual environment, you should install the libraries necessary for the FEDOT framework to work. In order to do this, go to the terminal console (blue square) and run the following command pip install .[extra] (red square).

    • For more details, take a look at the picture below.

      Step 3

  • Step 3. Manually installing libraries.

    • In order to use the