ex_ml_call_shap.py

shapvals = s.linearExplainer(
             table           = {"name" : 'hmeqTest','caslib':'public'},
             query           = {"name" : 'realtime','caslib':'public'},
             modelTable      = {"name" :"hmeqTestAstore",'caslib':'public'},
             modelTableType  = "ASTORE",
             predictedTarget = 'P_BAD1',
             seed            = 1234,
             preset          = "KERNELSHAP",
             inputs          = ['LOAN','MORTDUE','VALUE','YOJ','DEROG','DELINQ','CLAGE','NINQ','CLNO','DEBTINC','REASON', 'JOB','BAD'],
             nominals        = ['REASON', 'JOB','BAD']
            )
shap1 = shapvals['ParameterEstimates']
shap = shap1[['Variable','Estimate']][0:10]

ex_ml_cas_connect.py

s = CAS(host, port, username, password)
s.loadActionSet('autotune')
s.loadactionset('aStore')
s.loadactionset('decisionTree')
s.loadactionset("explainModel")
s.loadactionset('table')

ex_ml_dicttopd_convert.py

def dicttopd(datadict):
    for key in datadict:
        datadict[key] = [datadict[key]]
    return pd.DataFrame.from_dict(datadict)

ex_ml_export.py

s.table.promote(name="hmeqTest", caslib='public',target="hmeqTest",targetLib='public')
modelAstore = s.decisionTree.dtreeExportModel(modelTable = {"caslib":"public","name":"gradboosthmeqtest" }, 
                                        casOut = {"caslib":"public","name":'hmeqTestAstore','replace':True})

s.table.promote(name='hmeqTestAstore', caslib='public',target='hmeqTestAstore',targetLib='public')

ex_ml_hmeq.py

hmeq = pd.read_csv('hmeq.csv')
hmeq

ex_ml_hmeq_gb.py

result = s.autotune.tuneGradientBoostTree(
    trainOptions = {
        "table"   : {"name":'hmeqTest', 'caslib' : 'public'},
        "inputs"  : {'LOAN','MORTDUE','VALUE','YOJ','DEROG','DELINQ','CLAGE','NINQ','CLNO','DEBTINC','REASON', 'JOB'},
        "target"  : 'BAD',
        "nominal" : {'BAD','REASON', 'JOB'},
        "casout"  : {"name":"gradboosthmeqtest", "caslib":"public",'replace':True},
        "varImp" : True
    },
    tunerOptions={"seed":12345, "maxTime":60}
)

ex_ml_imports.py

!pip install swat
from swat import CAS, options
import pandas as pd
import numpy as np

ex_ml_obs_upload.py

samplepd = dicttopd(datadict)
s.upload(samplepd,casout={'name' : 'realtime', 'caslib' : 'public','replace' : True})

ex_ml_sample.py

datadict = {'LOAN':140,'MORTDUE':3000, 'VALUE':40000, 'REASON':'HomeImp','JOB':'Other','YOJ':12,
           'DEROG':0.0,'DELINQ':0.0, 'CLAGE':89,'NINQ':1.0, 'CLNO':10.0, 'DEBTINC':0.05} 

ex_ml_score.py

s.aStore.score(rstore = {"caslib":"public","name":"hmeqTestAstore"},
                    table = {"caslib":'public',"name":'realtime'},
                    out = {"caslib":'public',"name":'realscore', 'replace':True})
scoredData = s.CASTable(name='realscore',caslib='public')
datasetDict = scoredData.to_dict()
scores = pd.DataFrame(datasetDict, index=[0])
scores

ex_ml_score_func.py

def score(samplepd):
    s.upload(samplepd,casout={'name' : 'realtime', 'caslib' : 'public','replace' : True})
    s.aStore.score(rstore = {"caslib":"public","name":"hmeqTestAstore"},
                        table = {"caslib":'public',"name":'realtime'},
                        out = {"caslib":'public',"name":'realscore', 'replace':True})
    scoredData = s.CASTable(name='realscore',caslib='public')
    datasetDict = scoredData.to_dict()
    scores = pd.DataFrame(datasetDict, index=[0])
    return scores
    

ex_ml_upload.py

s.upload(hmeq,casout={'name' : 'hmeqTest', 'caslib' : 'public','replace' : True})

ex_ml_visualize.py

!pip install altair
import altair as alt
alt.Chart(shap).mark_bar().encode(
    x='Variable',
    y='Estimate'
)

ex_to_interpret.py

samplepd['BAD'] = scores.I_BAD.to_list()
s.upload(samplepd,casout={'name' : 'realtime', 'caslib' : 'public','replace' : True})