predictnow-client 0.0.9

A restful client library, designed to access predictnow restful api.

Usage:

from predictnow.pdapi import PredictNowClient import pandas as pd api_key = "%KeyProvidedToEachOfOurSubscriber" 

api_host = "http://%VMIP%"  # current makeshift api host username = "helloWorld" email = "[e-mail]"

client = PredictNowClient(api_host,api_key)

You will need to edit this input dataset file path and labelname!

file_path = 'my_amazing_features.xlsx' labelname = 'Next_day_strategy_return' import os

For classification problems

params = {'timeseries': 'yes', 'type': 'classification', 'feature_selection': 'shap', 'analysis': 'none', 'boost': 'gbdt', 'testsize': '0.2', 'weights': 'no', 'eda': 'yes', 'prob_calib': 'no', 'mode': 'train'}

For regression problems, suitable for CPO

params = {'timeseries': 'yes', 'type': 'regression', 'feature_selection': 'none', 'analysis': 'none', 'boost': 'gbdt', 'testsize': '0.2', 'weights': 'no', 'eda': 'yes', 'prob_calib': 'no', 'mode': 'train'}

response = client.create_model( username=username, model_name="test1", params=params )

from pandas import read_excel df = read_excel(file_path, engine="openpyxl")  # Same here df.name = "testdataframe"  # Optional, but recommended

########## TRAIN MODE ################################### response = client.train( model_name="test1", input_df=df, label=labelname, username=username, email=email)

status = client.getstatus(      username=username,     train_id=response["train_id"] )

if status["state"] == "COMPLETED":

response = client.getresult(
model_name="test1",
username=username,
)
predicted_targets_cv = pd.read_json(response.predicted_targets_cv)
print("predicted_targets_cv")
print(predicted_targets_cv)

predicted_targets_test = pd.read_json(response.predicted_targets_test)
print("predicted_targets_test")
print(predicted_targets_test)

if response.feature_importance:
	feature_importance = pd.read_json(response.feature_importance)
	print("feature_importance")
	print(feature_importance)


performance_metrics = pd.read_json(response.performance_metrics)
print("performance_metrics")
print(performance_metrics)

########## LIVE MODE ################################### if status["state"] == "COMPLETED": df = read_csv("example_input_live.csv") # live input data df.name = "myfirstpredictname"  # optional, but recommended # Making live predictions response = client.predict( model_name="test1", input_df=df,username=username, eda="yes", prob_calib=params["prob_calib"], ) y_pred = pd.read_json(response.labels) print("THE LABELS") print(y_pred)