This code finds the best algorithm from sklearn to forecast your numeric data
To getting started with pythonFirecastMachineLearning you just clone this repository in your project folder and import forecastingML in your file
git clone https://github.com/sebashc3712/pythonForecastMachineLearning
from ForecastingML import bestForecastModel
Then you have to call the function bestForecastingModel and save the model into a file .pkl
a,b=bestForecastModel(Dataset,Target)
joblib.dump(a, 'bestForecastModel.pkl')
bestForecastModel(Dataset, Target, MLP = False, MaxLayers=100, MaxDepth=10, neighbors=20)
- MLP (deafult False) when is True the neural network algorithm is activated. (This can take a while. Until 24 hrs depengin of the number of max layers and the numbers of rows of your dataset)
- MaxLayers (default 100). It only works if MLP is True. This parameter is the highest number of hidden layers for the algorithm
- MaxDepth (default 10). That is the maximum deep for the random forest and decision tree algorithms
- neighbors (default 20). That is the maximun neighbors for the isomap algorithm
You need to install sklearn and pandas. The easiest way to install these packages is to install Anaconda in your PC
https://www.anaconda.com/download/
MIT