modelcreator - AutoML package

This package contains a Machine which is meant to do the learning for you. It can automaticly create a fitting predictive model for given data.

Sample output

Testing:  Gradient Boosting Classifier
[########################################] | 100% Completed |  3.9s
Score: 0.9667

Testing:  Ada Boost Classifier
[########################################] | 100% Completed |  1.3s
Score: 0.9600

Testing:  Random Forest Classifier
[########################################] | 100% Completed |  5.0s
Score: 0.9600

Testing:  Balanced Random Forest Classifier
[########################################] | 100% Completed |  3.5s
Score: 0.9600

Testing:  SVC
[########################################] | 100% Completed |  1.2s
Score: 0.9667

Chosen model:  Gradient Boosting Classifier 0.9667

Params:
        min_samples_split: 2
        n_estimators: 100

Results saved to  output.csv

Table of Contents

1. Installation
2. Usage
   • CSV input
   • Pandas input
3. Saving model
4. Parameters
   • Machine
   • learn
   • learnFromDf
   • predict
   • predictFromDf
   • saveMachine
5. Development

Installation

To use the package run:

pip install modelcreator

Usage

The input may be either a path to a csv file or a pandas DataFrame object.

CSV path input

The library assumes that the last column of the training dataset contains the expected results. The dataset (both training and predictive) must be provided as a csv file.

If the results column contains text the Machine will do its best to learn to classify the data correctly. In case of a number inside, regression will be performed.

If the file contains headers you shall add header_in_csv=True parameter to the method.

Example 1 Iris

from modelcreator import Machine

# Create automl machine instance
machine = Machine()

# Train machine learning model
machine.learn('example-data/iris.csv')

# Predict the outcomes
machine.predict('example-data/iris-pred.csv', 'output.csv')

This example is also available in the example.py file. Consider trying it on your own.

Pandas input

But what to do if a result column is not the last in the given csv? It may be inconvenient to rewrite the whole csv just to swap the columns. Because of this problem Machine has learnFromDf and predictFromDf methods. The Df in method names stands for DataFrame from pandas module. This way you can handle reading the file by yourself.

Example 2 Titanic

from modelcreator import Machine
import pandas as pd

# Create DataFrame object from file
train = pd.read_csv("train.csv")

# Get features columns from DataFrame
X_train = train.drop(['Survived'], axis=1)

# And labels (results) column
y_train = train["Survived"].astype(str)

# Create the instance of Machine
machine = Machine()

# Train machine learning model
machine.learnFromDf(X_train, y_train, computation_level='advanced')

# Show parameters of the model
machine.showParams()

# Load test set from file
X_test = pd.read_csv("test.csv")

# Predict the labels
results = machine.predictFromDf(X_test)

# Save results to a new file
results.to_csv("results.csv")

Simple? That's right! Just note that we used astype(str) in order to treat data as classes, not numbers because the Titanic dataset used in the example above has values 0 and 1 in "Survived" column to indicate whether a person made it through the disaster.

Saving the model

If you want your model to avoid re-learning on the whole dataset just to make a simple prediction you can save the state of Machine to a file.

# Save Machine with a trained model to "machine.pkl"
machine.saveMachine('machine.pkl')

# Create a new machine based on a schema file
machine2 = Machine('machine.pkl')

Parameters

The Machine can be customized according to the use case. Check the parameters table:

Machine

learn

learnFromDf

predict

predictFromDf

saveMachine

Development

Have a feature idea or just want to help? Take a look at the issues tab!