Assisting library for the ML4CV tutorial based on scikit-learn.
It is recommended to use Python 3.6 in a virtual environment and install the latest stable versions of the dependencies. If not present, mlcv-tutorial will attempt to install them automatically.
mlcv-tutorial requires:
- numpy (>= 1.13.3)
- scipy (>= 0.19.1)
- scikit-learn (>=0.19.0)
- requests (>=2.14.2)
- matplotlib (>=2.0.2)
Create a virtual environment. If you use
pip
:python3 -m venv /path/to/new/virtual/environment_name
or if you use
conda
:conda create -n environment_name python=3.6 anaconda
Enter the virtual environment:
source activate environment_name
Install or upgrade the package:
pip install --upgrade git+https://github.com/johny-c/mlcv-tutorial.git
To exit the virtual environment:
source deactivate
Enter the virtual environment you created. Upgrade regularly to get the latest version. Open a python script, import the package and use it in your own work!
from mlcv.templates.base import Solution
class MyEstimator(Solution):
def __init__(param1=3, param2='gaussian'):
# Store the passed parameters in your estimator instance
self.param1 = param1
self.param2 = param2
def fit(X, y):
# Train your estimator on the training inputs X and training targets y
return self
def predict(X):
# Predict targets for the given testing inputs X.
return y_pred
def score(y_pred, y_true):
# Evaluate your model
return accuracy
Have a look at the examples directory for a complete use case.