class NNPTRegressionScript(object):
def __init__(self,
data_frame,
df_helper,
df_context,
spark,
prediction_narrative,
result_setter,
meta_parser,
mlEnvironment="sklearn"):
self._metaParser = meta_parser
self._prediction_narrative = prediction_narrative
self._result_setter = result_setter
self._data_frame = data_frame
self._dataframe_helper = df_helper
self._dataframe_context = df_context
self._spark = spark
self._model_summary = MLModelSummary()
self._score_summary = {}
self._slug = GLOBALSETTINGS.MODEL_SLUG_MAPPING[
"Neural Network (PyTorch)"]
self._analysisName = self._slug
self._dataframe_context.set_analysis_name(self._analysisName)
self._mlEnv = mlEnvironment
self._datasetName = CommonUtils.get_dataset_name(
self._dataframe_context.CSV_FILE)
self._completionStatus = self._dataframe_context.get_completion_status(
)
print(self._completionStatus, "initial completion status")
self._messageURL = self._dataframe_context.get_message_url()
self._scriptWeightDict = self._dataframe_context.get_ml_model_training_weight(
)
self._ignoreMsg = self._dataframe_context.get_message_ignore()
self._scriptStages = {
"initialization": {
"summary": "Initialized The Neural Network (PyTorch) Scripts",
"weight": 1
},
"training": {
"summary": "Neural Network (PyTorch) Training Started",
"weight": 2
},
"completion": {
"summary": "Neural Network (PyTorch) Training Finished",
"weight": 1
},
}
def Train(self):
st_global = time.time()
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"initialization",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
appType = self._dataframe_context.get_app_type()
algosToRun = self._dataframe_context.get_algorithms_to_run()
algoSetting = [
x for x in algosToRun if x.get_algorithm_slug() == self._slug
][0]
categorical_columns = self._dataframe_helper.get_string_columns()
uid_col = self._dataframe_context.get_uid_column()
if self._metaParser.check_column_isin_ignored_suggestion(uid_col):
categorical_columns = list(set(categorical_columns) - {uid_col})
allDateCols = self._dataframe_context.get_date_columns()
categorical_columns = list(set(categorical_columns) - set(allDateCols))
print("CATEGORICAL COLS - ", categorical_columns)
result_column = self._dataframe_context.get_result_column()
numerical_columns = self._dataframe_helper.get_numeric_columns()
numerical_columns = [
x for x in numerical_columns if x != result_column
]
model_path = self._dataframe_context.get_model_path()
if model_path.startswith("file"):
model_path = model_path[7:]
validationDict = self._dataframe_context.get_validation_dict()
print("model_path", model_path)
pipeline_filepath = "file://" + str(model_path) + "/" + str(
self._slug) + "/pipeline/"
model_filepath = "file://" + str(model_path) + "/" + str(
self._slug) + "/model"
pmml_filepath = "file://" + str(model_path) + "/" + str(
self._slug) + "/modelPmml"
df = self._data_frame
if self._mlEnv == "spark":
pass
elif self._mlEnv == "sklearn":
model_filepath = model_path + "/" + self._slug + "/model.pkl"
x_train, x_test, y_train, y_test = self._dataframe_helper.get_train_test_data(
)
x_train = MLUtils.create_dummy_columns(
x_train,
[x for x in categorical_columns if x != result_column])
x_test = MLUtils.create_dummy_columns(
x_test, [x for x in categorical_columns if x != result_column])
x_test = MLUtils.fill_missing_columns(x_test, x_train.columns,
result_column)
print("=" * 150)
print("X-Train Shape - ", x_train.shape)
print("Y-Train Shape - ", y_train.shape)
print("X-Test Shape - ", x_test.shape)
print("Y-Test Shape - ", y_test.shape)
print("~" * 50)
print("X-Train dtype - ", type(x_train))
print("Y-Train dtype - ", type(y_train))
print("X-Test dtype - ", type(x_test))
print("Y-Test dtype - ", type(y_test))
print("~" * 50)
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"training",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
st = time.time()
self._result_setter.set_hyper_parameter_results(self._slug, None)
evaluationMetricDict = algoSetting.get_evaluvation_metric(
Type="REGRESSION")
evaluationMetricDict = {
"name": GLOBALSETTINGS.REGRESSION_MODEL_EVALUATION_METRIC
}
evaluationMetricDict[
"displayName"] = GLOBALSETTINGS.SKLEARN_EVAL_METRIC_NAME_DISPLAY_MAP[
evaluationMetricDict["name"]]
x_train_tensored, y_train_tensored, x_test_tensored, y_test_tensored = PYTORCHUTILS.get_tensored_data(
x_train, y_train, x_test, y_test)
trainset = torch_data_utils.TensorDataset(x_train_tensored,
y_train_tensored)
testset = torch_data_utils.TensorDataset(x_test_tensored,
y_test_tensored)
nnptr_params = algoSetting.get_nnptr_params_dict()[0]
layers_for_network = PYTORCHUTILS.get_layers_for_network_module(
nnptr_params,
task_type="REGRESSION",
first_layer_units=x_train.shape[1])
# Use GPU if available
device = torch.device(
"cuda:0" if torch.cuda.is_available() else "cpu")
network = PyTorchNetwork(layers_for_network).to(device)
network.eval()
other_params_dict = PYTORCHUTILS.get_other_pytorch_params(
nnptr_params,
task_type="REGRESSION",
network_params=network.parameters())
print("~" * 50)
print("NNPTR-PARAMS - ", nnptr_params)
print("~" * 50)
print("OTHER-PARAMS-DICT - ", other_params_dict)
print("~" * 50)
print("NEURAL-NETWORK - ", network)
print("~" * 50)
criterion = other_params_dict["loss_criterion"]
n_epochs = other_params_dict["number_of_epochs"]
batch_size = other_params_dict["batch_size"]
optimizer = other_params_dict["optimizer"]
dataloader_params = {
"batch_size": batch_size,
"shuffle": True
# "num_workers":
}
train_loader = torch_data_utils.DataLoader(trainset,
**dataloader_params)
test_loader = torch_data_utils.DataLoader(testset,
**dataloader_params)
'''
Training the network;
Batchnormalization(num_features) should be equal to units_op for that layer in training config;
else --> RuntimeError('running_mean should contain 100 elements not 200',)
'''
for epoch in range(n_epochs):
batchwise_losses = []
average_loss = 0.0
for i, (inputs, labels) in enumerate(train_loader):
inputs = inputs.to(device)
labels = labels.to(device)
# Zero the parameter gradients
optimizer.zero_grad()
# Forward + backward + optimize
outputs = network(inputs.float())
loss = criterion(outputs, labels.float())
loss.backward()
optimizer.step()
average_loss += loss.item()
batchwise_losses.append(loss.item())
average_loss_per_epoch = old_div(average_loss, (i + 1))
print("+" * 80)
print("EPOCH - ", epoch)
print("BATCHWISE_LOSSES shape - ", len(batchwise_losses))
print("AVERAGE LOSS PER EPOCH - ", average_loss_per_epoch)
print("+" * 80)
trainingTime = time.time() - st
bestEstimator = network
outputs_x_test_tensored = network(x_test_tensored.float())
y_score_mid = outputs_x_test_tensored.tolist()
y_score = [x[0] for x in y_score_mid]
print("Y-SCORE - ", y_score)
print("Y-SCORE length - ", len(y_score))
y_prob = None
featureImportance = {}
objs = {
"trained_model": bestEstimator,
"actual": y_test,
"predicted": y_score,
"probability": y_prob,
"feature_importance": featureImportance,
"featureList": list(x_train.columns),
"labelMapping": {}
}
#featureImportance = objs["trained_model"].feature_importances_
#featuresArray = [(col_name, featureImportance[idx]) for idx, col_name in enumerate(x_train.columns)]
featuresArray = []
if not algoSetting.is_hyperparameter_tuning_enabled():
modelName = "M" + "0" * (GLOBALSETTINGS.MODEL_NAME_MAX_LENGTH -
1) + "1"
modelFilepathArr = model_filepath.split("/")[:-1]
modelFilepathArr.append(modelName + ".pt")
torch.save(objs["trained_model"], "/".join(modelFilepathArr))
#joblib.dump(objs["trained_model"],"/".join(modelFilepathArr))
runtime = round((time.time() - st), 2)
else:
runtime = round((time.time() - hyper_st), 2)
try:
modelPmmlPipeline = PMMLPipeline([("pretrained-estimator",
objs["trained_model"])])
modelPmmlPipeline.target_field = result_column
modelPmmlPipeline.active_fields = np.array(
[col for col in x_train.columns if col != result_column])
sklearn2pmml(modelPmmlPipeline, pmml_filepath, with_repr=True)
pmmlfile = open(pmml_filepath, "r")
pmmlText = pmmlfile.read()
pmmlfile.close()
self._result_setter.update_pmml_object({self._slug: pmmlText})
except:
pass
metrics = {}
metrics["r2"] = r2_score(y_test, y_score)
metrics["neg_mean_squared_error"] = mean_squared_error(
y_test, y_score)
metrics["neg_mean_absolute_error"] = mean_absolute_error(
y_test, y_score)
metrics["RMSE"] = sqrt(metrics["neg_mean_squared_error"])
metrics["explained_variance_score"] = explained_variance_score(
y_test, y_score)
transformed = pd.DataFrame({
"prediction": y_score,
result_column: y_test
})
print("TRANSFORMED PREDICTION TYPE - ",
type(transformed["prediction"]))
print(transformed["prediction"])
print("TRANSFORMED RESULT COL TYPE - ",
type(transformed[result_column]))
print(transformed[result_column])
transformed["difference"] = transformed[
result_column] - transformed["prediction"]
transformed["mape"] = old_div(
np.abs(transformed["difference"]) * 100,
transformed[result_column])
sampleData = None
nrows = transformed.shape[0]
if nrows > 100:
sampleData = transformed.sample(n=100, random_state=420)
else:
sampleData = transformed
print(sampleData.head())
if transformed["mape"].max() > 100:
GLOBALSETTINGS.MAPEBINS.append(transformed["mape"].max())
mapeCountArr = list(
pd.cut(transformed["mape"], GLOBALSETTINGS.MAPEBINS).
value_counts().to_dict().items())
GLOBALSETTINGS.MAPEBINS.pop(5)
else:
mapeCountArr = list(
pd.cut(transformed["mape"], GLOBALSETTINGS.MAPEBINS).
value_counts().to_dict().items())
mapeStatsArr = [(str(idx), dictObj) for idx, dictObj in enumerate(
sorted([{
"count": x[1],
"splitRange": (x[0].left, x[0].right)
} for x in mapeCountArr],
key=lambda x: x["splitRange"][0]))]
print(mapeStatsArr)
print(mapeCountArr)
predictionColSummary = transformed["prediction"].describe(
).to_dict()
quantileBins = [
predictionColSummary["min"], predictionColSummary["25%"],
predictionColSummary["50%"], predictionColSummary["75%"],
predictionColSummary["max"]
]
print(quantileBins)
quantileBins = sorted(list(set(quantileBins)))
transformed["quantileBinId"] = pd.cut(transformed["prediction"],
quantileBins)
quantileDf = transformed.groupby("quantileBinId").agg({
"prediction": [np.sum, np.mean, np.size]
}).reset_index()
quantileDf.columns = ["prediction", "sum", "mean", "count"]
print(quantileDf)
quantileArr = list(quantileDf.T.to_dict().items())
quantileSummaryArr = [(obj[0], {
"splitRange":
(obj[1]["prediction"].left, obj[1]["prediction"].right),
"count":
obj[1]["count"],
"mean":
obj[1]["mean"],
"sum":
obj[1]["sum"]
}) for obj in quantileArr]
print(quantileSummaryArr)
runtime = round((time.time() - st_global), 2)
self._model_summary.set_model_type("regression")
self._model_summary.set_algorithm_name("Neural Network (PyTorch)")
self._model_summary.set_algorithm_display_name(
"Neural Network (PyTorch)")
self._model_summary.set_slug(self._slug)
self._model_summary.set_training_time(runtime)
self._model_summary.set_training_time(trainingTime)
self._model_summary.set_target_variable(result_column)
self._model_summary.set_validation_method(
validationDict["displayName"])
self._model_summary.set_model_evaluation_metrics(metrics)
self._model_summary.set_model_params(nnptr_params)
self._model_summary.set_quantile_summary(quantileSummaryArr)
self._model_summary.set_mape_stats(mapeStatsArr)
self._model_summary.set_sample_data(sampleData.to_dict())
self._model_summary.set_feature_importance(featuresArray)
self._model_summary.set_feature_list(list(x_train.columns))
self._model_summary.set_model_mse(
metrics["neg_mean_squared_error"])
self._model_summary.set_model_mae(
metrics["neg_mean_absolute_error"])
self._model_summary.set_rmse(metrics["RMSE"])
self._model_summary.set_model_rsquared(metrics["r2"])
self._model_summary.set_model_exp_variance_score(
metrics["explained_variance_score"])
try:
pmml_filepath = str(model_path) + "/" + str(
self._slug) + "/traindeModel.pmml"
modelPmmlPipeline = PMMLPipeline([("pretrained-estimator",
objs["trained_model"])])
modelPmmlPipeline.target_field = result_column
modelPmmlPipeline.active_fields = np.array(
[col for col in x_train.columns if col != result_column])
sklearn2pmml(modelPmmlPipeline, pmml_filepath, with_repr=True)
pmmlfile = open(pmml_filepath, "r")
pmmlText = pmmlfile.read()
pmmlfile.close()
self._result_setter.update_pmml_object({self._slug: pmmlText})
except:
pass
if algoSetting.is_hyperparameter_tuning_enabled():
modelDropDownObj = {
"name": self._model_summary.get_algorithm_name(),
"evaluationMetricValue": metrics[evaluationMetricDict["name"]],
"evaluationMetricName": evaluationMetricDict["name"],
"slug": self._model_summary.get_slug(),
"Model Id": modelName
}
modelSummaryJson = {
"dropdown": modelDropDownObj,
"levelcount": self._model_summary.get_level_counts(),
"modelFeatureList": self._model_summary.get_feature_list(),
"levelMapping": self._model_summary.get_level_map_dict(),
"slug": self._model_summary.get_slug(),
"name": self._model_summary.get_algorithm_name()
}
else:
modelDropDownObj = {
"name": self._model_summary.get_algorithm_name(),
"evaluationMetricValue": metrics[evaluationMetricDict["name"]],
"evaluationMetricName": evaluationMetricDict["name"],
"slug": self._model_summary.get_slug(),
"Model Id": modelName
}
modelSummaryJson = {
"dropdown": modelDropDownObj,
"levelcount": self._model_summary.get_level_counts(),
"modelFeatureList": self._model_summary.get_feature_list(),
"levelMapping": self._model_summary.get_level_map_dict(),
"slug": self._model_summary.get_slug(),
"name": self._model_summary.get_algorithm_name()
}
modelmanagement_ = nnptr_params
self._model_management = MLModelSummary()
if algoSetting.is_hyperparameter_tuning_enabled():
pass
else:
self._model_management.set_layer_info(
data=modelmanagement_['hidden_layer_info'])
self._model_management.set_loss_function(
data=modelmanagement_['loss'])
self._model_management.set_optimizer(
data=modelmanagement_['optimizer'])
self._model_management.set_batch_size(
data=modelmanagement_['batch_size'])
self._model_management.set_no_epochs(
data=modelmanagement_['number_of_epochs'])
# self._model_management.set_model_evaluation_metrics(data=modelmanagement_['metrics'])
self._model_management.set_job_type(
self._dataframe_context.get_job_name()) #Project name
self._model_management.set_training_status(
data="completed") # training status
self._model_management.set_no_of_independent_variables(
data=x_train) #no of independent varables
self._model_management.set_training_time(runtime) # run time
self._model_management.set_rmse(metrics["RMSE"])
self._model_management.set_algorithm_name(
"Neural Network (TensorFlow)") #algorithm name
self._model_management.set_validation_method(
str(validationDict["displayName"]) + "(" +
str(validationDict["value"]) + ")") #validation method
self._model_management.set_target_variable(
result_column) #target column name
self._model_management.set_creation_date(data=str(
datetime.now().strftime('%b %d ,%Y %H:%M '))) #creation date
self._model_management.set_datasetName(self._datasetName)
modelManagementSummaryJson = [
["Project Name",
self._model_management.get_job_type()],
["Algorithm",
self._model_management.get_algorithm_name()],
["Training Status",
self._model_management.get_training_status()],
["RMSE", self._model_management.get_rmse()],
["RunTime", self._model_management.get_training_time()],
#["Owner",None],
["Created On",
self._model_management.get_creation_date()]
]
if algoSetting.is_hyperparameter_tuning_enabled():
modelManagementModelSettingsJson = []
else:
modelManagementModelSettingsJson = [
["Training Dataset",
self._model_management.get_datasetName()],
[
"Target Column",
self._model_management.get_target_variable()
],
[
"Number Of Independent Variables",
self._model_management.get_no_of_independent_variables()
], ["Algorithm",
self._model_management.get_algorithm_name()],
[
"Model Validation",
self._model_management.get_validation_method()
],
["batch_size",
str(self._model_management.get_batch_size())],
["Loss", self._model_management.get_loss_function()],
["Optimizer",
self._model_management.get_optimizer()],
["Epochs", self._model_management.get_no_epochs()],
[
"Metrics",
self._model_management.get_model_evaluation_metrics()
]
]
for i in modelmanagement_["hidden_layer_info"]:
string = ""
key = str(modelmanagement_["hidden_layer_info"][i]
["layer"]) + " " + str(i) + ":"
for j in modelmanagement_["hidden_layer_info"][i]:
string = string + str(j) + ":" + str(
modelmanagement_["hidden_layer_info"][i][j]) + ", "
modelManagementModelSettingsJson.append([key, string])
print(modelManagementModelSettingsJson)
nnptrCards = [
json.loads(CommonUtils.convert_python_object_to_json(cardObj)) for
cardObj in MLUtils.create_model_summary_cards(self._model_summary)
]
nnptrPerformanceCards = [
json.loads(CommonUtils.convert_python_object_to_json(cardObj))
for cardObj in MLUtils.create_model_management_cards_regression(
self._model_summary)
]
nnptrOverviewCards = [
json.loads(CommonUtils.convert_python_object_to_json(cardObj))
for cardObj in MLUtils.create_model_management_card_overview(
self._model_management, modelManagementSummaryJson,
modelManagementModelSettingsJson)
]
nnptrDeploymentCards = [
json.loads(CommonUtils.convert_python_object_to_json(cardObj))
for cardObj in MLUtils.create_model_management_deploy_empty_card()
]
nnptr_Overview_Node = NarrativesTree()
nnptr_Overview_Node.set_name("Overview")
nnptr_Performance_Node = NarrativesTree()
nnptr_Performance_Node.set_name("Performance")
nnptr_Deployment_Node = NarrativesTree()
nnptr_Deployment_Node.set_name("Deployment")
for card in nnptrOverviewCards:
nnptr_Overview_Node.add_a_card(card)
for card in nnptrPerformanceCards:
nnptr_Performance_Node.add_a_card(card)
for card in nnptrDeploymentCards:
nnptr_Deployment_Node.add_a_card(card)
for card in nnptrCards:
self._prediction_narrative.add_a_card(card)
self._result_setter.set_model_summary({
"Neural Network (PyTorch)":
json.loads(
CommonUtils.convert_python_object_to_json(self._model_summary))
})
self._result_setter.set_nnptr_regression_model_summary(
modelSummaryJson)
self._result_setter.set_nnptr_cards(nnptrCards)
self._result_setter.set_nnptr_nodes([
nnptr_Overview_Node, nnptr_Performance_Node, nnptr_Deployment_Node
])
self._result_setter.set_nnptr_fail_card({
"Algorithm_Name": "Neural Network (PyTorch)",
"Success": "True"
})
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"completion",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
def Predict(self):
self._scriptWeightDict = self._dataframe_context.get_ml_model_prediction_weight(
)
self._scriptStages = {
"initialization": {
"summary": "Initialized The Neural Network (PyTorch) Scripts",
"weight": 2
},
"predictionStart": {
"summary": "Neural Network (PyTorch) Prediction Started",
"weight": 2
},
"predictionFinished": {
"summary": "Neural Network (PyTorch) Prediction Finished",
"weight": 6
}
}
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"initialization",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
SQLctx = SQLContext(sparkContext=self._spark.sparkContext,
sparkSession=self._spark)
dataSanity = True
categorical_columns = self._dataframe_helper.get_string_columns()
uid_col = self._dataframe_context.get_uid_column()
if self._metaParser.check_column_isin_ignored_suggestion(uid_col):
categorical_columns = list(set(categorical_columns) - {uid_col})
allDateCols = self._dataframe_context.get_date_columns()
categorical_columns = list(set(categorical_columns) - set(allDateCols))
numerical_columns = self._dataframe_helper.get_numeric_columns()
result_column = self._dataframe_context.get_result_column()
test_data_path = self._dataframe_context.get_input_file()
if self._mlEnv == "spark":
pass
elif self._mlEnv == "sklearn":
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"predictionStart",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
score_data_path = self._dataframe_context.get_score_path(
) + "/data.csv"
trained_model_path = "file://" + self._dataframe_context.get_model_path(
)
trained_model_path += "/" + self._dataframe_context.get_model_for_scoring(
) + ".pt"
print("trained_model_path", trained_model_path)
print("score_data_path", score_data_path)
if trained_model_path.startswith("file"):
trained_model_path = trained_model_path[7:]
#trained_model = joblib.load(trained_model_path)
trained_model = torch.load(trained_model_path,
map_location=torch.device('cpu'))
model_columns = self._dataframe_context.get_model_features()
print("model_columns", model_columns)
try:
df = self._data_frame.toPandas()
except:
df = self._data_frame
# pandas_df = MLUtils.factorize_columns(df,[x for x in categorical_columns if x != result_column])
pandas_df = MLUtils.create_dummy_columns(
df, [x for x in categorical_columns if x != result_column])
pandas_df = MLUtils.fill_missing_columns(pandas_df, model_columns,
result_column)
if uid_col:
pandas_df = pandas_df[[
x for x in pandas_df.columns if x != uid_col
]]
test_df = np.stack(
[pandas_df[col].values for col in pandas_df.columns], 1)
tensored_test_df = torch.tensor(test_df, dtype=torch.float)
outputs_test_df_tensored = trained_model(tensored_test_df.float())
y_score_mid = outputs_test_df_tensored.tolist()
y_score = [x[0] for x in y_score_mid]
scoreKpiArray = MLUtils.get_scored_data_summary(y_score)
kpiCard = NormalCard()
kpiCardData = [KpiData(data=x) for x in scoreKpiArray]
kpiCard.set_card_data(kpiCardData)
kpiCard.set_cente_alignment(True)
print(CommonUtils.convert_python_object_to_json(kpiCard))
self._result_setter.set_kpi_card_regression_score(kpiCard)
pandas_df[result_column] = y_score
df[result_column] = y_score
df.to_csv(score_data_path, header=True, index=False)
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"predictionFinished",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
print("STARTING Measure ANALYSIS ...")
columns_to_keep = []
columns_to_drop = []
columns_to_keep = self._dataframe_context.get_score_consider_columns(
)
if len(columns_to_keep) > 0:
columns_to_drop = list(set(df.columns) - set(columns_to_keep))
else:
columns_to_drop += ["predicted_probability"]
columns_to_drop = [
x for x in columns_to_drop
if x in df.columns and x != result_column
]
print("columns_to_drop", columns_to_drop)
pandas_scored_df = df[list(set(columns_to_keep + [result_column]))]
spark_scored_df = SQLctx.createDataFrame(pandas_scored_df)
# spark_scored_df.write.csv(score_data_path+"/data",mode="overwrite",header=True)
# TODO update metadata for the newly created dataframe
self._dataframe_context.update_consider_columns(columns_to_keep)
print(spark_scored_df.printSchema())
df_helper = DataFrameHelper(spark_scored_df, self._dataframe_context,
self._metaParser)
df_helper.set_params()
df = df_helper.get_data_frame()
# self._dataframe_context.set_dont_send_message(True)
try:
fs = time.time()
descr_stats_obj = DescriptiveStatsScript(
df,
df_helper,
self._dataframe_context,
self._result_setter,
self._spark,
self._prediction_narrative,
scriptWeight=self._scriptWeightDict,
analysisName="Descriptive analysis")
descr_stats_obj.Run()
print("DescriptiveStats Analysis Done in ",
time.time() - fs, " seconds.")
except:
print("Frequency Analysis Failed ")
# try:
# fs = time.time()
# df_helper.fill_na_dimension_nulls()
# df = df_helper.get_data_frame()
# dt_reg = DecisionTreeRegressionScript(df, df_helper, self._dataframe_context, self._result_setter, self._spark,self._prediction_narrative,self._metaParser,scriptWeight=self._scriptWeightDict,analysisName="Predictive modeling")
# dt_reg.Run()
# print "DecisionTrees Analysis Done in ", time.time() - fs, " seconds."
# except:
# print "DTREE FAILED"
try:
fs = time.time()
two_way_obj = TwoWayAnovaScript(
df,
df_helper,
self._dataframe_context,
self._result_setter,
self._spark,
self._prediction_narrative,
self._metaParser,
scriptWeight=self._scriptWeightDict,
analysisName="Measure vs. Dimension")
two_way_obj.Run()
print("OneWayAnova Analysis Done in ",
time.time() - fs, " seconds.")
except:
print("Anova Analysis Failed")
class GeneralizedLinearRegressionModelScript:
def __init__(self, data_frame, df_helper,df_context, spark, prediction_narrative, result_setter,meta_parser):
self._metaParser = meta_parser
self._prediction_narrative = prediction_narrative
self._result_setter = result_setter
self._data_frame = data_frame
self._dataframe_helper = df_helper
self._dataframe_context = df_context
self._spark = spark
self._model_summary = MLModelSummary()
self._score_summary = {}
self._slug = GLOBALSETTINGS.MODEL_SLUG_MAPPING["generalizedlinearregression"]
def Train(self):
st_global = time.time()
algosToRun = self._dataframe_context.get_algorithms_to_run()
algoSetting = filter(lambda x:x["algorithmSlug"]==GLOBALSETTINGS.MODEL_SLUG_MAPPING["generalizedlinearregression"],algosToRun)[0]
categorical_columns = self._dataframe_helper.get_string_columns()
uid_col = self._dataframe_context.get_uid_column()
if self._metaParser.check_column_isin_ignored_suggestion(uid_col):
categorical_columns = list(set(categorical_columns) - {uid_col})
allDateCols = self._dataframe_context.get_date_columns()
categorical_columns = list(set(categorical_columns)-set(allDateCols))
print categorical_columns
result_column = self._dataframe_context.get_result_column()
numerical_columns = self._dataframe_helper.get_numeric_columns()
numerical_columns = [x for x in numerical_columns if x != result_column]
model_path = self._dataframe_context.get_model_path()
if model_path.startswith("file"):
model_path = model_path[7:]
validationDict = self._dataframe_context.get_validation_dict()
print "model_path",model_path
pipeline_filepath = "file://"+str(model_path)+"/"+str(self._slug)+"/pipeline/"
model_filepath = "file://"+str(model_path)+"/"+str(self._slug)+"/model"
pmml_filepath = "file://"+str(model_path)+"/"+str(self._slug)+"/modelPmml"
df = self._data_frame
pipeline = MLUtils.create_pyspark_ml_pipeline(numerical_columns,categorical_columns,result_column,algoType="regression")
pipelineModel = pipeline.fit(df)
indexed = pipelineModel.transform(df)
featureMapping = sorted((attr["idx"], attr["name"]) for attr in (chain(*indexed.schema["features"].metadata["ml_attr"]["attrs"].values())))
# print indexed.select([result_column,"features"]).show(5)
MLUtils.save_pipeline_or_model(pipelineModel,pipeline_filepath)
glinr = GeneralizedLinearRegression(labelCol=result_column, featuresCol='features',predictionCol="prediction")
if validationDict["name"] == "kFold":
defaultSplit = GLOBALSETTINGS.DEFAULT_VALIDATION_OBJECT["value"]
numFold = int(validationDict["value"])
if numFold == 0:
numFold = 3
trainingData,validationData = indexed.randomSplit([defaultSplit,1-defaultSplit], seed=12345)
paramGrid = ParamGridBuilder()\
.addGrid(glinr.regParam, [0.1, 0.01]) \
.addGrid(glinr.fitIntercept, [False, True])\
.build()
crossval = CrossValidator(estimator=glinr,
estimatorParamMaps=paramGrid,
evaluator=RegressionEvaluator(predictionCol="prediction", labelCol=result_column),
numFolds=numFold)
st = time.time()
cvModel = crossval.fit(indexed)
trainingTime = time.time()-st
print "cvModel training takes",trainingTime
bestModel = cvModel.bestModel
elif validationDict["name"] == "trainAndtest":
trainingData,validationData = indexed.randomSplit([float(validationDict["value"]),1-float(validationDict["value"])], seed=12345)
st = time.time()
fit = glinr.fit(trainingData)
trainingTime = time.time()-st
print "time to train",trainingTime
bestModel = fit
print bestModel.explainParams()
print bestModel.extractParamMap()
print bestModel.params
print 'Best Param (regParam): ', bestModel._java_obj.getRegParam()
print 'Best Param (MaxIter): ', bestModel._java_obj.getMaxIter()
# modelPmmlPipeline = PMMLPipeline([
# ("pretrained-estimator", objs["trained_model"])
# ])
# try:
# modelPmmlPipeline.target_field = result_column
# modelPmmlPipeline.active_fields = np.array([col for col in x_train.columns if col != result_column])
# sklearn2pmml(modelPmmlPipeline, pmml_filepath, with_repr = True)
# pmmlfile = open(pmml_filepath,"r")
# pmmlText = pmmlfile.read()
# pmmlfile.close()
# self._result_setter.update_pmml_object({self._slug:pmmlText})
# except:
# pass
coefficientsArray = [(name, bestModel.coefficients[idx]) for idx, name in featureMapping]
MLUtils.save_pipeline_or_model(bestModel,model_filepath)
transformed = bestModel.transform(validationData)
transformed = transformed.withColumn(result_column,transformed[result_column].cast(DoubleType()))
transformed = transformed.select([result_column,"prediction",transformed[result_column]-transformed["prediction"]])
transformed = transformed.withColumnRenamed(transformed.columns[-1],"difference")
transformed = transformed.select([result_column,"prediction","difference",FN.abs(transformed["difference"])*100/transformed[result_column]])
transformed = transformed.withColumnRenamed(transformed.columns[-1],"mape")
sampleData = None
nrows = transformed.count()
if nrows > 100:
sampleData = transformed.sample(False, float(100)/nrows, seed=420)
else:
sampleData = transformed
print sampleData.show()
evaluator = RegressionEvaluator(predictionCol="prediction",labelCol=result_column)
metrics = {}
metrics["r2"] = evaluator.evaluate(transformed,{evaluator.metricName: "r2"})
metrics["rmse"] = evaluator.evaluate(transformed,{evaluator.metricName: "rmse"})
metrics["mse"] = evaluator.evaluate(transformed,{evaluator.metricName: "mse"})
metrics["mae"] = evaluator.evaluate(transformed,{evaluator.metricName: "mae"})
runtime = round((time.time() - st_global),2)
# print transformed.count()
mapeDf = transformed.select("mape")
# print mapeDf.show()
mapeStats = MLUtils.get_mape_stats(mapeDf,"mape")
mapeStatsArr = mapeStats.items()
mapeStatsArr = sorted(mapeStatsArr,key=lambda x:int(x[0]))
# print mapeStatsArr
quantileDf = transformed.select("prediction")
# print quantileDf.show()
quantileSummaryDict = MLUtils.get_quantile_summary(quantileDf,"prediction")
quantileSummaryArr = quantileSummaryDict.items()
quantileSummaryArr = sorted(quantileSummaryArr,key=lambda x:int(x[0]))
# print quantileSummaryArr
self._model_summary.set_model_type("regression")
self._model_summary.set_algorithm_name("Generalized Linear Regression")
self._model_summary.set_algorithm_display_name("Generalized Linear Regression")
self._model_summary.set_slug(self._slug)
self._model_summary.set_training_time(runtime)
self._model_summary.set_training_time(trainingTime)
self._model_summary.set_target_variable(result_column)
self._model_summary.set_validation_method(validationDict["displayName"])
self._model_summary.set_model_evaluation_metrics(metrics)
self._model_summary.set_model_params(bestEstimator.get_params())
self._model_summary.set_quantile_summary(quantileSummaryArr)
self._model_summary.set_mape_stats(mapeStatsArr)
self._model_summary.set_sample_data(sampleData.toPandas().to_dict())
self._model_summary.set_coefficinets_array(coefficientsArray)
self._model_summary.set_feature_list(list(x_train.columns))
# print CommonUtils.convert_python_object_to_json(self._model_summary)
modelSummaryJson = {
"dropdown":{
"name":self._model_summary.get_algorithm_name(),
"accuracy":CommonUtils.round_sig(self._model_summary.get_model_evaluation_metrics()["r2"]),
"slug":self._model_summary.get_slug()
},
"levelcount":self._model_summary.get_level_counts(),
"modelFeatureList":self._model_summary.get_feature_list(),
"levelMapping":self._model_summary.get_level_map_dict()
}
glinrCards = [json.loads(CommonUtils.convert_python_object_to_json(cardObj)) for cardObj in MLUtils.create_model_summary_cards(self._model_summary)]
for card in glinrCards:
self._prediction_narrative.add_a_card(card)
self._result_setter.set_model_summary({"generalizedlinearregression":json.loads(CommonUtils.convert_python_object_to_json(self._model_summary))})
self._result_setter.set_generalized_linear_regression_model_summary(modelSummaryJson)
self._result_setter.set_glinr_cards(glinrCards)
def Predict(self):
self._scriptWeightDict = self._dataframe_context.get_ml_model_prediction_weight()
self._scriptStages = {
"initialization":{
"summary":"Initialized the Generalized Linear Regression Scripts",
"weight":2
},
"predictionStart":{
"summary":"Generalized Linear Regression Model Prediction Started",
"weight":2
},
"predictionFinished":{
"summary":"Generalized Linear Regression Model Prediction Finished",
"weight":6
}
}
CommonUtils.create_update_and_save_progress_message(self._dataframe_context,self._scriptWeightDict,self._scriptStages,self._slug,"initialization","info",display=True,emptyBin=False,customMsg=None,weightKey="total")
SQLctx = SQLContext(sparkContext=self._spark.sparkContext, sparkSession=self._spark)
dataSanity = True
categorical_columns = self._dataframe_helper.get_string_columns()
uid_col = self._dataframe_context.get_uid_column()
if self._metaParser.check_column_isin_ignored_suggestion(uid_col):
categorical_columns = list(set(categorical_columns) - {uid_col})
allDateCols = self._dataframe_context.get_date_columns()
categorical_columns = list(set(categorical_columns)-set(allDateCols))
numerical_columns = self._dataframe_helper.get_numeric_columns()
result_column = self._dataframe_context.get_result_column()
test_data_path = self._dataframe_context.get_input_file()
CommonUtils.create_update_and_save_progress_message(self._dataframe_context,self._scriptWeightDict,self._scriptStages,self._slug,"predictionStart","info",display=True,emptyBin=False,customMsg=None,weightKey="total")
test_data_path = self._dataframe_context.get_input_file()
score_data_path = self._dataframe_context.get_score_path()+"/data.csv"
trained_model_path = "file://" + self._dataframe_context.get_model_path()
trained_model_path += "/model"
pipeline_path = "/".join(trained_model_path.split("/")[:-1])+"/pipeline"
print "trained_model_path",trained_model_path
print "pipeline_path",pipeline_path
print "score_data_path",score_data_path
pipelineModel = MLUtils.load_pipeline(pipeline_path)
trained_model = MLUtils.load_generalized_linear_regresssion_pyspark_model(trained_model_path)
df = self._data_frame
indexed = pipelineModel.transform(df)
transformed = trained_model.transform(indexed)
if result_column in transformed.columns:
transformed = transformed.withColumnRenamed(result_column,"originalLabel")
transformed = transformed.withColumnRenamed("prediction",result_column)
pandas_scored_df = transformed.select(list(set(self._data_frame.columns+[result_column]))).toPandas()
if score_data_path.startswith("file"):
score_data_path = score_data_path[7:]
pandas_scored_df.to_csv(score_data_path,header=True,index=False)
CommonUtils.create_update_and_save_progress_message(self._dataframe_context,self._scriptWeightDict,self._scriptStages,self._slug,"predictionFinished","info",display=True,emptyBin=False,customMsg=None,weightKey="total")
print "STARTING Measure ANALYSIS ..."
columns_to_keep = []
columns_to_drop = []
columns_to_keep = self._dataframe_context.get_score_consider_columns()
if len(columns_to_keep) > 0:
columns_to_drop = list(set(df.columns)-set(columns_to_keep))
else:
columns_to_drop += ["predicted_probability"]
columns_to_drop = [x for x in columns_to_drop if x in df.columns and x != result_column]
print "columns_to_drop",columns_to_drop
spark_scored_df = transformed.select(list(set(columns_to_keep+[result_column])))
df_helper = DataFrameHelper(spark_scored_df, self._dataframe_context,self._metaParser)
df_helper.set_params()
df = df_helper.get_data_frame()
# self._dataframe_context.set_dont_send_message(True)
try:
fs = time.time()
descr_stats_obj = DescriptiveStatsScript(df, df_helper, self._dataframe_context, self._result_setter, self._spark,self._prediction_narrative,scriptWeight=self._scriptWeightDict,analysisName="Descriptive analysis")
descr_stats_obj.Run()
print "DescriptiveStats Analysis Done in ", time.time() - fs, " seconds."
except:
print "Frequency Analysis Failed "
try:
fs = time.time()
df_helper.fill_na_dimension_nulls()
df = df_helper.get_data_frame()
dt_reg = DecisionTreeRegressionScript(df, df_helper, self._dataframe_context, self._result_setter, self._spark,self._prediction_narrative,self._metaParser,scriptWeight=self._scriptWeightDict,analysisName="Predictive modeling")
dt_reg.Run()
print "DecisionTrees Analysis Done in ", time.time() - fs, " seconds."
except:
print "DTREE FAILED"
try:
fs = time.time()
two_way_obj = TwoWayAnovaScript(df, df_helper, self._dataframe_context, self._result_setter, self._spark,self._prediction_narrative,self._metaParser,scriptWeight=self._scriptWeightDict,analysisName="Measure vs. Dimension")
two_way_obj.Run()
print "OneWayAnova Analysis Done in ", time.time() - fs, " seconds."
except:
print "Anova Analysis Failed"
class TensorFlowRegScript(object):
def __init__(self,
data_frame,
df_helper,
df_context,
spark,
prediction_narrative,
result_setter,
meta_parser,
mlEnvironment="sklearn"):
self._metaParser = meta_parser
self._prediction_narrative = prediction_narrative
self._result_setter = result_setter
self._data_frame = data_frame
self._dataframe_helper = df_helper
self._dataframe_context = df_context
self._spark = spark
self._model_summary = MLModelSummary()
self._score_summary = {}
self._slug = GLOBALSETTINGS.MODEL_SLUG_MAPPING[
"Neural Network (TensorFlow)"]
self._analysisName = "Neural Network (TensorFlow)"
self._dataframe_context.set_analysis_name(self._analysisName)
self._mlEnv = mlEnvironment
self._datasetName = CommonUtils.get_dataset_name(
self._dataframe_context.CSV_FILE)
self._completionStatus = self._dataframe_context.get_completion_status(
)
print(self._completionStatus, "initial completion status")
self._messageURL = self._dataframe_context.get_message_url()
self._scriptWeightDict = self._dataframe_context.get_ml_model_training_weight(
)
self._ignoreMsg = self._dataframe_context.get_message_ignore()
self._scriptStages = {
"initialization": {
"summary":
"Initialized The Neural Network (TensorFlow) Regression Scripts",
"weight": 1
},
"training": {
"summary":
"Neural Network (TensorFlow) Regression Model Training Started",
"weight": 2
},
"completion": {
"summary":
"Neural Network (TensorFlow) Regression Model Training Finished",
"weight": 1
},
}
def Train(self):
st_global = time.time()
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"initialization",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
appType = self._dataframe_context.get_app_type()
algosToRun = self._dataframe_context.get_algorithms_to_run()
algoSetting = [
x for x in algosToRun if x.get_algorithm_slug() == self._slug
][0]
categorical_columns = self._dataframe_helper.get_string_columns()
uid_col = self._dataframe_context.get_uid_column()
if self._metaParser.check_column_isin_ignored_suggestion(uid_col):
categorical_columns = list(set(categorical_columns) - {uid_col})
allDateCols = self._dataframe_context.get_date_columns()
categorical_columns = list(set(categorical_columns) - set(allDateCols))
print(categorical_columns)
result_column = self._dataframe_context.get_result_column()
numerical_columns = self._dataframe_helper.get_numeric_columns()
numerical_columns = [
x for x in numerical_columns if x != result_column
]
model_path = self._dataframe_context.get_model_path()
if model_path.startswith("file"):
model_path = model_path[7:]
validationDict = self._dataframe_context.get_validation_dict()
print("model_path", model_path)
pipeline_filepath = "file://" + str(model_path) + "/" + str(
self._slug) + "/pipeline/"
model_filepath = "file://" + str(model_path) + "/" + str(
self._slug) + "/model"
pmml_filepath = "file://" + str(model_path) + "/" + str(
self._slug) + "/modelPmml"
df = self._data_frame
if self._mlEnv == "spark":
pass
elif self._mlEnv == "sklearn":
model_filepath = model_path + "/" + self._slug + "/model.pkl"
x_train, x_test, y_train, y_test = self._dataframe_helper.get_train_test_data(
)
x_train = MLUtils.create_dummy_columns(
x_train,
[x for x in categorical_columns if x != result_column])
x_test = MLUtils.create_dummy_columns(
x_test, [x for x in categorical_columns if x != result_column])
x_test = MLUtils.fill_missing_columns(x_test, x_train.columns,
result_column)
st = time.time()
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"training",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
if algoSetting.is_hyperparameter_tuning_enabled():
pass
else:
self._result_setter.set_hyper_parameter_results(
self._slug, None)
evaluationMetricDict = algoSetting.get_evaluvation_metric(
Type="Regression")
evaluationMetricDict[
"displayName"] = GLOBALSETTINGS.SKLEARN_EVAL_METRIC_NAME_DISPLAY_MAP[
evaluationMetricDict["name"]]
params_tf = algoSetting.get_tf_params_dict()
algoParams = algoSetting.get_params_dict()
algoParams = {k: v for k, v in list(algoParams.items())}
model = tf.keras.models.Sequential()
first_layer_flag = True
for i in range(len(list(
params_tf['hidden_layer_info'].keys()))):
if params_tf['hidden_layer_info'][str(
i)]["layer"] == "Dense":
if first_layer_flag:
model.add(
tf.keras.layers.Dense(
params_tf['hidden_layer_info'][str(
i)]["units"],
activation=params_tf['hidden_layer_info'][
str(i)]["activation"],
input_shape=(len(x_train.columns), ),
use_bias=params_tf['hidden_layer_info'][
str(i)]["use_bias"],
kernel_initializer=params_tf[
'hidden_layer_info'][str(
i)]["kernel_initializer"],
bias_initializer=params_tf[
'hidden_layer_info'][str(
i)]["bias_initializer"],
kernel_regularizer=params_tf[
'hidden_layer_info'][str(
i)]["kernel_regularizer"],
bias_regularizer=params_tf[
'hidden_layer_info'][str(
i)]["bias_regularizer"],
activity_regularizer=params_tf[
'hidden_layer_info'][str(
i)]["activity_regularizer"],
kernel_constraint=params_tf[
'hidden_layer_info'][str(
i)]["kernel_constraint"],
bias_constraint=params_tf[
'hidden_layer_info'][str(
i)]["bias_constraint"]))
try:
if params_tf['hidden_layer_info'][str(
i)]["batch_normalization"] == "True":
model.add(
tf.keras.layers.BatchNormalization())
except:
print(
"BATCH_NORM_FAILED ##########################"
)
pass
first_layer_flag = False
else:
model.add(
tf.keras.layers.Dense(
params_tf['hidden_layer_info'][str(
i)]["units"],
activation=params_tf['hidden_layer_info'][
str(i)]["activation"],
use_bias=params_tf['hidden_layer_info'][
str(i)]["use_bias"],
kernel_initializer=params_tf[
'hidden_layer_info'][str(
i)]["kernel_initializer"],
bias_initializer=params_tf[
'hidden_layer_info'][str(
i)]["bias_initializer"],
kernel_regularizer=params_tf[
'hidden_layer_info'][str(
i)]["kernel_regularizer"],
bias_regularizer=params_tf[
'hidden_layer_info'][str(
i)]["bias_regularizer"],
activity_regularizer=params_tf[
'hidden_layer_info'][str(
i)]["activity_regularizer"],
kernel_constraint=params_tf[
'hidden_layer_info'][str(
i)]["kernel_constraint"],
bias_constraint=params_tf[
'hidden_layer_info'][str(
i)]["bias_constraint"]))
try:
if params_tf['hidden_layer_info'][str(
i)]["batch_normalization"] == "True":
model.add(
tf.keras.layers.BatchNormalization())
except:
print(
"BATCH_NORM_FAILED ##########################"
)
pass
elif params_tf['hidden_layer_info'][str(
i)]["layer"] == "Dropout":
model.add(
tf.keras.layers.Dropout(
float(params_tf['hidden_layer_info'][str(i)]
["rate"])))
elif params_tf['hidden_layer_info'][str(
i)]["layer"] == "Lambda":
if params_tf['hidden_layer_info'][str(
i)]["lambda"] == "Addition":
model.add(
tf.keras.layers.Lambda(lambda x: x + int(
params_tf['hidden_layer_info'][str(i)][
"units"])))
if params_tf['hidden_layer_info'][str(
i)]["lambda"] == "Multiplication":
model.add(
tf.keras.layers.Lambda(lambda x: x * int(
params_tf['hidden_layer_info'][str(i)][
"units"])))
if params_tf['hidden_layer_info'][str(
i)]["lambda"] == "Subtraction":
model.add(
tf.keras.layers.Lambda(lambda x: x - int(
params_tf['hidden_layer_info'][str(i)][
"units"])))
if params_tf['hidden_layer_info'][str(
i)]["lambda"] == "Division":
model.add(
tf.keras.layers.Lambda(lambda x: old_div(
x,
int(params_tf['hidden_layer_info'][str(i)][
"units"]))))
model.compile(optimizer=algoParams["optimizer"],
loss=algoParams["loss"],
metrics=[algoParams['metrics']])
model.fit(x_train,
y_train,
epochs=algoParams["number_of_epochs"],
verbose=1,
batch_size=algoParams["batch_size"])
bestEstimator = model
print(model.summary())
trainingTime = time.time() - st
y_score = bestEstimator.predict(x_test)
y_score = list(y_score.flatten())
try:
y_prob = bestEstimator.predict_proba(x_test)
except:
y_prob = [0] * len(y_score)
featureImportance = {}
objs = {
"trained_model": bestEstimator,
"actual": y_test,
"predicted": y_score,
"probability": y_prob,
"feature_importance": featureImportance,
"featureList": list(x_train.columns),
"labelMapping": {}
}
#featureImportance = objs["trained_model"].feature_importances_
#featuresArray = [(col_name, featureImportance[idx]) for idx, col_name in enumerate(x_train.columns)]
featuresArray = []
if not algoSetting.is_hyperparameter_tuning_enabled():
modelName = "M" + "0" * (GLOBALSETTINGS.MODEL_NAME_MAX_LENGTH -
1) + "1"
modelFilepathArr = model_filepath.split("/")[:-1]
modelFilepathArr.append(modelName + ".h5")
objs["trained_model"].save("/".join(modelFilepathArr))
#joblib.dump(objs["trained_model"],"/".join(modelFilepathArr))
metrics = {}
metrics["r2"] = r2_score(y_test, y_score)
metrics["neg_mean_squared_error"] = mean_squared_error(
y_test, y_score)
metrics["neg_mean_absolute_error"] = mean_absolute_error(
y_test, y_score)
metrics["RMSE"] = sqrt(metrics["neg_mean_squared_error"])
metrics["explained_variance_score"] = explained_variance_score(
y_test, y_score)
transformed = pd.DataFrame({
"prediction": y_score,
result_column: y_test
})
transformed["difference"] = transformed[
result_column] - transformed["prediction"]
transformed["mape"] = old_div(
np.abs(transformed["difference"]) * 100,
transformed[result_column])
sampleData = None
nrows = transformed.shape[0]
if nrows > 100:
sampleData = transformed.sample(n=100, random_state=420)
else:
sampleData = transformed
print(sampleData.head())
if transformed["mape"].max() > 100:
GLOBALSETTINGS.MAPEBINS.append(transformed["mape"].max())
mapeCountArr = list(
pd.cut(transformed["mape"], GLOBALSETTINGS.MAPEBINS).
value_counts().to_dict().items())
GLOBALSETTINGS.MAPEBINS.pop(5)
else:
mapeCountArr = list(
pd.cut(transformed["mape"], GLOBALSETTINGS.MAPEBINS).
value_counts().to_dict().items())
mapeStatsArr = [(str(idx), dictObj) for idx, dictObj in enumerate(
sorted([{
"count": x[1],
"splitRange": (x[0].left, x[0].right)
} for x in mapeCountArr],
key=lambda x: x["splitRange"][0]))]
print(mapeStatsArr)
print(mapeCountArr)
predictionColSummary = transformed["prediction"].describe(
).to_dict()
quantileBins = [
predictionColSummary["min"], predictionColSummary["25%"],
predictionColSummary["50%"], predictionColSummary["75%"],
predictionColSummary["max"]
]
print(quantileBins)
quantileBins = sorted(list(set(quantileBins)))
transformed["quantileBinId"] = pd.cut(transformed["prediction"],
quantileBins)
quantileDf = transformed.groupby("quantileBinId").agg({
"prediction": [np.sum, np.mean, np.size]
}).reset_index()
quantileDf.columns = ["prediction", "sum", "mean", "count"]
print(quantileDf)
quantileArr = list(quantileDf.T.to_dict().items())
quantileSummaryArr = [(obj[0], {
"splitRange":
(obj[1]["prediction"].left, obj[1]["prediction"].right),
"count":
obj[1]["count"],
"mean":
obj[1]["mean"],
"sum":
obj[1]["sum"]
}) for obj in quantileArr]
print(quantileSummaryArr)
runtime = round((time.time() - st_global), 2)
self._model_summary.set_model_type("regression")
self._model_summary.set_algorithm_name(
"Neural Network (TensorFlow)")
self._model_summary.set_algorithm_display_name(
"Neural Network (TensorFlow)")
self._model_summary.set_slug(self._slug)
self._model_summary.set_training_time(runtime)
self._model_summary.set_training_time(trainingTime)
self._model_summary.set_target_variable(result_column)
self._model_summary.set_validation_method(
validationDict["displayName"])
self._model_summary.set_model_evaluation_metrics(metrics)
self._model_summary.set_model_params(params_tf)
self._model_summary.set_quantile_summary(quantileSummaryArr)
self._model_summary.set_mape_stats(mapeStatsArr)
self._model_summary.set_sample_data(sampleData.to_dict())
self._model_summary.set_feature_importance(featuresArray)
self._model_summary.set_feature_list(list(x_train.columns))
self._model_summary.set_model_mse(
metrics["neg_mean_squared_error"])
self._model_summary.set_model_mae(
metrics["neg_mean_absolute_error"])
self._model_summary.set_rmse(metrics["RMSE"])
self._model_summary.set_model_rsquared(metrics["r2"])
self._model_summary.set_model_exp_variance_score(
metrics["explained_variance_score"])
try:
pmml_filepath = str(model_path) + "/" + str(
self._slug) + "/traindeModel.pmml"
modelPmmlPipeline = PMMLPipeline([("pretrained-estimator",
objs["trained_model"])])
modelPmmlPipeline.target_field = result_column
modelPmmlPipeline.active_fields = np.array(
[col for col in x_train.columns if col != result_column])
sklearn2pmml(modelPmmlPipeline, pmml_filepath, with_repr=True)
pmmlfile = open(pmml_filepath, "r")
pmmlText = pmmlfile.read()
pmmlfile.close()
self._result_setter.update_pmml_object({self._slug: pmmlText})
except:
pass
if algoSetting.is_hyperparameter_tuning_enabled():
modelDropDownObj = {
"name": self._model_summary.get_algorithm_name(),
"evaluationMetricValue": metrics[evaluationMetricDict["name"]],
"evaluationMetricName": evaluationMetricDict["name"],
"slug": self._model_summary.get_slug(),
"Model Id": modelName
}
modelSummaryJson = {
"dropdown": modelDropDownObj,
"levelcount": self._model_summary.get_level_counts(),
"modelFeatureList": self._model_summary.get_feature_list(),
"levelMapping": self._model_summary.get_level_map_dict(),
"slug": self._model_summary.get_slug(),
"name": self._model_summary.get_algorithm_name()
}
else:
modelDropDownObj = {
"name": self._model_summary.get_algorithm_name(),
"evaluationMetricValue": metrics[evaluationMetricDict["name"]],
"evaluationMetricName": evaluationMetricDict["name"],
"slug": self._model_summary.get_slug(),
"Model Id": modelName
}
modelSummaryJson = {
"dropdown": modelDropDownObj,
"levelcount": self._model_summary.get_level_counts(),
"modelFeatureList": self._model_summary.get_feature_list(),
"levelMapping": self._model_summary.get_level_map_dict(),
"slug": self._model_summary.get_slug(),
"name": self._model_summary.get_algorithm_name()
}
modelmanagement_ = params_tf
modelmanagement_.update(algoParams)
self._model_management = MLModelSummary()
if algoSetting.is_hyperparameter_tuning_enabled():
pass
else:
self._model_management.set_layer_info(
data=modelmanagement_['hidden_layer_info'])
self._model_management.set_loss_function(
data=modelmanagement_['loss'])
self._model_management.set_optimizer(
data=modelmanagement_['optimizer'])
self._model_management.set_batch_size(
data=modelmanagement_['batch_size'])
self._model_management.set_no_epochs(
data=modelmanagement_['number_of_epochs'])
self._model_management.set_model_evaluation_metrics(
data=modelmanagement_['metrics'])
self._model_management.set_job_type(
self._dataframe_context.get_job_name()) #Project name
self._model_management.set_training_status(
data="completed") # training status
self._model_management.set_no_of_independent_variables(
data=x_train) #no of independent varables
self._model_management.set_training_time(runtime) # run time
self._model_management.set_rmse(metrics["RMSE"])
self._model_management.set_algorithm_name(
"Neural Network (TensorFlow)") #algorithm name
self._model_management.set_validation_method(
str(validationDict["displayName"]) + "(" +
str(validationDict["value"]) + ")") #validation method
self._model_management.set_target_variable(
result_column) #target column name
self._model_management.set_creation_date(data=str(
datetime.now().strftime('%b %d ,%Y %H:%M '))) #creation date
self._model_management.set_datasetName(self._datasetName)
modelManagementSummaryJson = [
["Project Name",
self._model_management.get_job_type()],
["Algorithm",
self._model_management.get_algorithm_name()],
["Training Status",
self._model_management.get_training_status()],
["RMSE", self._model_management.get_rmse()],
["RunTime", self._model_management.get_training_time()],
#["Owner",None],
["Created On",
self._model_management.get_creation_date()]
]
if algoSetting.is_hyperparameter_tuning_enabled():
modelManagementModelSettingsJson = []
else:
modelManagementModelSettingsJson = [
["Training Dataset",
self._model_management.get_datasetName()],
[
"Target Column",
self._model_management.get_target_variable()
],
[
"Number Of Independent Variables",
self._model_management.get_no_of_independent_variables()
], ["Algorithm",
self._model_management.get_algorithm_name()],
[
"Model Validation",
self._model_management.get_validation_method()
],
["batch_size",
str(self._model_management.get_batch_size())],
["Loss", self._model_management.get_loss_function()],
["Optimizer",
self._model_management.get_optimizer()],
["Epochs", self._model_management.get_no_epochs()],
[
"Metrics",
self._model_management.get_model_evaluation_metrics()
]
]
for i in range(
len(list(modelmanagement_['hidden_layer_info'].keys()))):
string = ""
key = "layer No-" + str(i) + "-" + str(
modelmanagement_["hidden_layer_info"][str(i)]["layer"] +
"-")
for j in modelmanagement_["hidden_layer_info"][str(i)]:
modelManagementModelSettingsJson.append([
key + j + ":",
modelmanagement_["hidden_layer_info"][str(i)][j]
])
print(modelManagementModelSettingsJson)
tfregCards = [
json.loads(CommonUtils.convert_python_object_to_json(cardObj)) for
cardObj in MLUtils.create_model_summary_cards(self._model_summary)
]
tfregPerformanceCards = [
json.loads(CommonUtils.convert_python_object_to_json(cardObj))
for cardObj in MLUtils.create_model_management_cards_regression(
self._model_summary)
]
tfregOverviewCards = [
json.loads(CommonUtils.convert_python_object_to_json(cardObj))
for cardObj in MLUtils.create_model_management_card_overview(
self._model_management, modelManagementSummaryJson,
modelManagementModelSettingsJson)
]
tfregDeploymentCards = [
json.loads(CommonUtils.convert_python_object_to_json(cardObj))
for cardObj in MLUtils.create_model_management_deploy_empty_card()
]
TFReg_Overview_Node = NarrativesTree()
TFReg_Overview_Node.set_name("Overview")
TFReg_Performance_Node = NarrativesTree()
TFReg_Performance_Node.set_name("Performance")
TFReg_Deployment_Node = NarrativesTree()
TFReg_Deployment_Node.set_name("Deployment")
for card in tfregOverviewCards:
TFReg_Overview_Node.add_a_card(card)
for card in tfregPerformanceCards:
TFReg_Performance_Node.add_a_card(card)
for card in tfregDeploymentCards:
TFReg_Deployment_Node.add_a_card(card)
for card in tfregCards:
self._prediction_narrative.add_a_card(card)
self._result_setter.set_model_summary({
"Neural Network (TensorFlow)":
json.loads(
CommonUtils.convert_python_object_to_json(self._model_summary))
})
self._result_setter.set_tfreg_regression_model_summart(
modelSummaryJson)
self._result_setter.set_tfreg_cards(tfregCards)
self._result_setter.set_tfreg_nodes([
TFReg_Overview_Node, TFReg_Performance_Node, TFReg_Deployment_Node
])
self._result_setter.set_tfreg_fail_card({
"Algorithm_Name": "Neural Network (TensorFlow)",
"Success": "True"
})
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"completion",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
def Predict(self):
self._scriptWeightDict = self._dataframe_context.get_ml_model_prediction_weight(
)
self._scriptStages = {
"initialization": {
"summary":
"Initialized The Neural Network (TensorFlow) Regression Scripts",
"weight": 2
},
"predictionStart": {
"summary":
"Neural Network (TensorFlow) Regression Model Prediction Started",
"weight": 2
},
"predictionFinished": {
"summary":
"Neural Network (TensorFlow) Regression Model Prediction Finished",
"weight": 6
}
}
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"initialization",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
SQLctx = SQLContext(sparkContext=self._spark.sparkContext,
sparkSession=self._spark)
dataSanity = True
categorical_columns = self._dataframe_helper.get_string_columns()
uid_col = self._dataframe_context.get_uid_column()
if self._metaParser.check_column_isin_ignored_suggestion(uid_col):
categorical_columns = list(set(categorical_columns) - {uid_col})
allDateCols = self._dataframe_context.get_date_columns()
categorical_columns = list(set(categorical_columns) - set(allDateCols))
numerical_columns = self._dataframe_helper.get_numeric_columns()
result_column = self._dataframe_context.get_result_column()
test_data_path = self._dataframe_context.get_input_file()
if self._mlEnv == "spark":
pass
elif self._mlEnv == "sklearn":
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"predictionStart",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
score_data_path = self._dataframe_context.get_score_path(
) + "/data.csv"
trained_model_path = "file://" + self._dataframe_context.get_model_path(
)
trained_model_path += "/" + self._dataframe_context.get_model_for_scoring(
) + ".h5"
print("trained_model_path", trained_model_path)
print("score_data_path", score_data_path)
if trained_model_path.startswith("file"):
trained_model_path = trained_model_path[7:]
#trained_model = joblib.load(trained_model_path)
trained_model = tf.keras.models.load_model(trained_model_path)
model_columns = self._dataframe_context.get_model_features()
print("model_columns", model_columns)
df = self._data_frame.toPandas()
# pandas_df = MLUtils.factorize_columns(df,[x for x in categorical_columns if x != result_column])
pandas_df = MLUtils.create_dummy_columns(
df, [x for x in categorical_columns if x != result_column])
pandas_df = MLUtils.fill_missing_columns(pandas_df, model_columns,
result_column)
if uid_col:
pandas_df = pandas_df[[
x for x in pandas_df.columns if x != uid_col
]]
y_score = trained_model.predict(pandas_df)
y_score = list(y_score.flatten())
scoreKpiArray = MLUtils.get_scored_data_summary(y_score)
kpiCard = NormalCard()
kpiCardData = [KpiData(data=x) for x in scoreKpiArray]
kpiCard.set_card_data(kpiCardData)
kpiCard.set_cente_alignment(True)
print(CommonUtils.convert_python_object_to_json(kpiCard))
self._result_setter.set_kpi_card_regression_score(kpiCard)
pandas_df[result_column] = y_score
df[result_column] = y_score
df.to_csv(score_data_path, header=True, index=False)
CommonUtils.create_update_and_save_progress_message(
self._dataframe_context,
self._scriptWeightDict,
self._scriptStages,
self._slug,
"predictionFinished",
"info",
display=True,
emptyBin=False,
customMsg=None,
weightKey="total")
print("STARTING Measure ANALYSIS ...")
columns_to_keep = []
columns_to_drop = []
columns_to_keep = self._dataframe_context.get_score_consider_columns(
)
if len(columns_to_keep) > 0:
columns_to_drop = list(set(df.columns) - set(columns_to_keep))
else:
columns_to_drop += ["predicted_probability"]
columns_to_drop = [
x for x in columns_to_drop
if x in df.columns and x != result_column
]
print("columns_to_drop", columns_to_drop)
pandas_scored_df = df[list(set(columns_to_keep + [result_column]))]
spark_scored_df = SQLctx.createDataFrame(pandas_scored_df)
# spark_scored_df.write.csv(score_data_path+"/data",mode="overwrite",header=True)
# TODO update metadata for the newly created dataframe
self._dataframe_context.update_consider_columns(columns_to_keep)
print(spark_scored_df.printSchema())
df_helper = DataFrameHelper(spark_scored_df, self._dataframe_context,
self._metaParser)
df_helper.set_params()
df = df_helper.get_data_frame()
# self._dataframe_context.set_dont_send_message(True)
try:
fs = time.time()
descr_stats_obj = DescriptiveStatsScript(
df,
df_helper,
self._dataframe_context,
self._result_setter,
self._spark,
self._prediction_narrative,
scriptWeight=self._scriptWeightDict,
analysisName="Descriptive analysis")
descr_stats_obj.Run()
print("DescriptiveStats Analysis Done in ",
time.time() - fs, " seconds.")
except:
print("Frequency Analysis Failed ")
# try:
# fs = time.time()
# df_helper.fill_na_dimension_nulls()
# df = df_helper.get_data_frame()
# dt_reg = DecisionTreeRegressionScript(df, df_helper, self._dataframe_context, self._result_setter, self._spark,self._prediction_narrative,self._metaParser,scriptWeight=self._scriptWeightDict,analysisName="Predictive modeling")
# dt_reg.Run()
# print "DecisionTrees Analysis Done in ", time.time() - fs, " seconds."
# except:
# print "DTREE FAILED"
try:
fs = time.time()
two_way_obj = TwoWayAnovaScript(
df,
df_helper,
self._dataframe_context,
self._result_setter,
self._spark,
self._prediction_narrative,
self._metaParser,
scriptWeight=self._scriptWeightDict,
analysisName="Measure vs. Dimension")
two_way_obj.Run()
print("OneWayAnova Analysis Done in ",
time.time() - fs, " seconds.")
except:
print("Anova Analysis Failed")