maadsbml

**Multi-Agent Accelerator for Data Science: Batch AutoML (MAADSBML)

Revolutionizing Data Science with Artificial Intelligence

Overview

*MAADSBML combines Artificial Intelligence, Docker, Machine Learning. It automates the machine learning process, and finds the BEST algorithm for your data. It also produces a very detailed PDF report. MAADSBML is integrated with Docker Container.

This library allows users to harness the power of agent-based computing using hundreds of advanced linear and non-linear algorithms.

FOR MORE INFORMATION GO TO: https://maadsbml.readthedocs.io/en/latest/

Compatibility - Python 3.8 or greater - Minimal Python skills needed

Copyright

• Author: Sebastian Maurice, PhD

Installation

• At the command prompt write: pip install maadsbml
  • This assumes you have Downloaded Python and installed it on your computer.
  • You will also need to pull the MAADSBML Docker container: maadsdocker/maads-batch-automl-otics

Syntax

• There are literally two lines of code you need to write to train your data and make predictions:

Main functions:

• hypertraining Executes hundreds of agents, running hundreds of advanced algorithms and completes in minutes. A master agent then chooses the BEST algorithm that best models your data.
• hyperpredictions After training, make high quality predictions - takes less than half a second (about ~100 milliseconds). Users can also generate predictions using non-python code such as JAVA.
• algodescription Get detailed information on the optimal algorithm found during hypertraining
• abort Abort the training process.
• rundemo To run canned demo of the system to see how it works.
• finddistribution Finds the best distribution for your continuous data

First import the Python library.

import maadsbml

1. maadsbml.hypertraining(host,port,filename,dependentvariable,removeoutliers=0,hasseasonality=0,summer='6,7,8',winter='11,12,1,2',shoulder='3,4,5,9,10', trainingpercentage=70,shuffle=0,deepanalysis=0,username='admin',timeout=1200,company='otics',password='123',email='[e-mail]',usereverseproxy=0, microserviceid='',maadstoken='123',mode=0)

Parameters:

host : string, required

• This is the IP address of the running Docker container - it is usually http://localhost

port : int, required

• This is the TRAINING PORT in the container. The default is port==5595

filename : string, required

• This is the raw data file in csv format - Note this file is stored on your host machine - the DOCKER container needs to be mapped to this volume using -v

dependentvariable : string, required

• This is the dependent variable in your csv file.

removeoutliers : int, optional, 1 or 0

• If 1, then outliers will be removed from your data. If 0, no outliers are removed.

hasseasonality : int, optional, 1 or 0

• If 1, then your data will be modeled for seasonality: Winter, Summer, Shoulder. If 0, then your data will not be modeled for seasonality. If modeling for seasonality, ensure you have enough data points that covers the seasons, usually 1 year of data.

summer : string, optional

• Definition for summer months. This can be changed.

winter : string, optional

• Definition for winter months. This can be changed.

shoulder : string, optional

• Definition for shoulder months. This can be changed.

trainingpercentage : int, optional, Default=70

• This is the split percentage between Training and Test data sets. It is defaulted to 70 (70% for training, 30% test).

shuffle : number, 0 or 1, optional

• Indicates whether to shuffle the training dataset or not, default=0.

deepanalysis : int, optional

• This will force MAADSBML to perform deeper analysis on your data. This could take 30-40 minutes. Set to 1 for deepanalysis, 0 for no deep analysis.

username : string, optional

• This identifies a user. You may want to change this if multiple users are running the same file.

company : string, optional

• This identifies your company. You may want to change this for the Report.

timeout : int, optional

• You can increase this if you receive a timeout error before the training is taking too long. The setting is in seconds.

password : string, optional

• leave as is

email : string, optional

• leave as is

usereverseproxy : int, optional

• leave as is

microserviceid : string, optional

• leave as is if not using a pass through service.

mode : int, optional

• leave as is

maadstoken : string, optional

• leave as is

Returns: string JSON buffer, with the algorithm key (PKEY) and other details:

• PKEY: : This is the key to the BEST algorithm and must be used when making predictions.

2. maadsbml.hyperpredictions(pkey,theinputdata,host,port,username,algoname='',seasonname='',usereverseproxy=0,microserviceid='',password='123',company='otics', email='[e-mail]',maadstoken='123')

Parameters:

pkey : string, required

• This is the PKEY you received from the hypertraining function.

theinputdata : string, required

• These are the Xs for your model: For example if my model had 3 Xs then inputdata='5/21/2010,-14.3,-32.0,-12.0', with the first entry as Date: Date must be in the format: MM/DD/YYYY

host : string, required

• This is the IP address of the running Docker container - it is usually http://localhost

port : int, required

• This is the PREDICTION PORT in the container. The default is port==5495 (or 5595)

username : string, required

• The username you used in the hypertraining functions. Default is admin.

algoname : string, optional

• Enter the name of the algorithm to use, this can be retrieved from the hypertraining function. If this is empty, the BEST algorithm will be used by default.

seasonname : string, optional

• Enter the season to use (winter,summer,shoulder), this can be retrieved from the hypertraining function. If this is empty, the default season is used.

usereverseproxy : int, optional

• leave as is

microserviceid : int, optional

• leave as is

password : string, optional

• leave as is

company : string, optional

• change for reporting.

email : string, optional

• leave as is

maadstoken : string, optional

• leave as is

Returns: string buffer containing the prediction, and other details.

3. maadsbml.abort(host,port=10000)

Parameters:

host : string, required

• This is the IP address of the Docker container: http://localhost

port : string, optional

• Port is fixed at 10000

Returns: Abort will shutdown and re-start your system.

4. maadsbml.rundemo(host,port,demotype=1,timeout=1200,usereverseproxy=0,microserviceid='')

Parameters:

host : string, required

• This is the IP address of the Docker container: http://localhost

port : string, required

• This is the TRAININGPORT, it is usually 5595.

demotype : int, required

• If demotype is 1, then a regression models is run; if demotype is 0 then a classification model is run.

timeout : int, optional

• The connection timeout between Python and the container, in seconds

usereverseproxy : int, optional

• leave as is

microserviceid : string, optional

• leave as is

Returns: null

5. maadsbml.algodescription(host,port,pkey,timeout=300,usereverseproxy=0,microserviceid='')

Parameters:

host : string, required

• This is the IP address of the Docker container: http://localhost

port : string, required

• This is the TRAININGPORT, it is usually 5595.

pkey : string, required

• This is the PKEY from hypertraining.

timeout : int, optional

• The connection timeout between Python and the container, in seconds

usereverseproxy : int, optional

• leave as is

microserviceid : string, optional

• leave as is

Returns: null

6. maadsbml.finddistribution(filename,varname,dataarray=[],folderpath='',imgname='distimage',common=1,topdist=5)

Parameters:

filename : string, required

• Filename containing the raw data. This must be a CSV file.

varname : string, required

• Name of the variable for your data.

dataarray : array_like, optional

• Numpy array. Rather than pass a filename, you can pass in an array.

folderpath : string, optional

• Folder path to store the output data, and distribution image file.

imgname : string, optional

• Name of the image and json data.

common : int, optional

• If set to 1, this will apply common distributions to your data.

  If Set to 0, it will iterate through roughly 80 distributions.

topdist : int, optional

• The number of the TOP distributions to print.

Returns: status,dist dataframe,name of best distribution,all JSON data