|
|
|
@ -359,7 +359,7 @@ def class_feature_importance(X, Y, feature_importances): |
|
|
|
|
|
|
|
|
|
|
|
#GridSearch = mem.cache(GridSearch) |
|
|
|
#GridSearch = mem.cache(GridSearch) |
|
|
|
|
|
|
|
|
|
|
|
def InitializeEnsemble(): |
|
|
|
def Preprocessing(): |
|
|
|
global resultsList |
|
|
|
global resultsList |
|
|
|
df_cv_results_classifiersList = [] |
|
|
|
df_cv_results_classifiersList = [] |
|
|
|
parametersList = [] |
|
|
|
parametersList = [] |
|
|
|
@ -378,48 +378,107 @@ def InitializeEnsemble(): |
|
|
|
FeatureImportance = pd.concat(FeatureImportanceList, ignore_index=True, sort=False) |
|
|
|
FeatureImportance = pd.concat(FeatureImportanceList, ignore_index=True, sort=False) |
|
|
|
PerClassMetrics = pd.concat(PerClassMetricsList, ignore_index=True, sort=False) |
|
|
|
PerClassMetrics = pd.concat(PerClassMetricsList, ignore_index=True, sort=False) |
|
|
|
FeatureAccuracy = pd.concat(FeatureAccuracyList, ignore_index=True, sort=False) |
|
|
|
FeatureAccuracy = pd.concat(FeatureAccuracyList, ignore_index=True, sort=False) |
|
|
|
|
|
|
|
global factors |
|
|
|
|
|
|
|
factors = [1,1,1,1,1,1] |
|
|
|
global scoring |
|
|
|
global scoring |
|
|
|
NumberofscoringMetrics = len(scoring) |
|
|
|
NumberofscoringMetrics = len(scoring) |
|
|
|
|
|
|
|
global df_cv_results_classifiers_metrics |
|
|
|
del df_cv_results_classifiers['params'] |
|
|
|
del df_cv_results_classifiers['params'] |
|
|
|
df_cv_results_classifiers_metrics = df_cv_results_classifiers.copy() |
|
|
|
df_cv_results_classifiers_metrics = df_cv_results_classifiers.copy() |
|
|
|
df_cv_results_classifiers_metrics = df_cv_results_classifiers_metrics.ix[:, 0:NumberofscoringMetrics+1] |
|
|
|
|
|
|
|
del df_cv_results_classifiers_metrics['mean_fit_time'] |
|
|
|
del df_cv_results_classifiers_metrics['mean_fit_time'] |
|
|
|
del df_cv_results_classifiers_metrics['mean_score_time'] |
|
|
|
del df_cv_results_classifiers_metrics['mean_score_time'] |
|
|
|
|
|
|
|
df_cv_results_classifiers_metrics = df_cv_results_classifiers_metrics.ix[:, 0:NumberofscoringMetrics] |
|
|
|
|
|
|
|
return [parameters,FeatureImportance,PerClassMetrics,FeatureAccuracy,df_cv_results_classifiers_metrics] |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def sumPerMetric(factors): |
|
|
|
sumPerClassifier = [] |
|
|
|
sumPerClassifier = [] |
|
|
|
|
|
|
|
global df_cv_results_classifiers_metrics |
|
|
|
for index, row in df_cv_results_classifiers_metrics.iterrows(): |
|
|
|
for index, row in df_cv_results_classifiers_metrics.iterrows(): |
|
|
|
rowSum = 0 |
|
|
|
rowSum = 0 |
|
|
|
for elements in row: |
|
|
|
global scoring |
|
|
|
rowSum = elements + rowSum |
|
|
|
lengthFactors = len(scoring) |
|
|
|
sumPerClassifier.append(rowSum) |
|
|
|
for loop,elements in enumerate(row): |
|
|
|
|
|
|
|
lengthFactors = lengthFactors - 1 + factors[loop] |
|
|
|
|
|
|
|
rowSum = elements*factors[loop] + rowSum |
|
|
|
|
|
|
|
sumPerClassifier.append(rowSum/lengthFactors) |
|
|
|
|
|
|
|
return sumPerClassifier |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# Retrieve data from client |
|
|
|
|
|
|
|
@cross_origin(origin='localhost',headers=['Content-Type','Authorization']) |
|
|
|
|
|
|
|
@app.route('/data/factors', methods=["GET", "POST"]) |
|
|
|
|
|
|
|
def RetrieveFactors(): |
|
|
|
|
|
|
|
Factors = request.get_data().decode('utf8').replace("'", '"') |
|
|
|
|
|
|
|
FactorsInt = json.loads(Factors) |
|
|
|
|
|
|
|
global sumPerClassifierSel |
|
|
|
|
|
|
|
global ModelSpaceMDSNew |
|
|
|
|
|
|
|
global ModelSpaceTSNENew |
|
|
|
|
|
|
|
sumPerClassifierSel = [] |
|
|
|
|
|
|
|
ModelSpaceMDSNew = [] |
|
|
|
|
|
|
|
ModelSpaceTSNENew = [] |
|
|
|
|
|
|
|
preProcessResults = [] |
|
|
|
|
|
|
|
preProcessResults = Preprocessing() |
|
|
|
|
|
|
|
XClassifiers = preProcessResults[4] |
|
|
|
|
|
|
|
flagLocal = 0 |
|
|
|
|
|
|
|
countRemovals = 0 |
|
|
|
|
|
|
|
for l,el in enumerate(FactorsInt['Factors']): |
|
|
|
|
|
|
|
if el is 0: |
|
|
|
|
|
|
|
XClassifiers.drop(XClassifiers.columns[[l-countRemovals]], axis=1, inplace=True) |
|
|
|
|
|
|
|
counter = countRemovals + 1 |
|
|
|
|
|
|
|
flagLocal = 1 |
|
|
|
|
|
|
|
if flagLocal is 1: |
|
|
|
|
|
|
|
ModelSpaceMDSNew = FunMDS(XClassifiers) |
|
|
|
|
|
|
|
ModelSpaceTSNENew = FunTsne(XClassifiers) |
|
|
|
|
|
|
|
ModelSpaceTSNENew = ModelSpaceTSNENew.tolist() |
|
|
|
|
|
|
|
sumPerClassifierSel = sumPerMetric(FactorsInt['Factors']) |
|
|
|
|
|
|
|
return 'Everything Okay' |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@app.route('/data/UpdateOverv', methods=["GET", "POST"]) |
|
|
|
|
|
|
|
def UpdateOverview(): |
|
|
|
|
|
|
|
global sumPerClassifierSel |
|
|
|
|
|
|
|
global ModelSpaceMDSNew |
|
|
|
|
|
|
|
global ModelSpaceTSNENew |
|
|
|
|
|
|
|
ResultsUpdateOverview = [] |
|
|
|
|
|
|
|
ResultsUpdateOverview.append(sumPerClassifierSel) |
|
|
|
|
|
|
|
ResultsUpdateOverview.append(ModelSpaceMDSNew) |
|
|
|
|
|
|
|
ResultsUpdateOverview.append(ModelSpaceTSNENew) |
|
|
|
|
|
|
|
response = { |
|
|
|
|
|
|
|
'Results': ResultsUpdateOverview |
|
|
|
|
|
|
|
} |
|
|
|
|
|
|
|
return jsonify(response) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def InitializeEnsemble(): |
|
|
|
|
|
|
|
preProcessResults = [] |
|
|
|
|
|
|
|
preProcessResults = Preprocessing() |
|
|
|
|
|
|
|
sumPerClassifier = sumPerMetric(factors) |
|
|
|
mergedPredList = zip(*yPredictProb) |
|
|
|
mergedPredList = zip(*yPredictProb) |
|
|
|
mergedPredListListForm = [] |
|
|
|
mergedPredListListForm = [] |
|
|
|
for el in mergedPredList: |
|
|
|
for el in mergedPredList: |
|
|
|
mergedPredListListForm.append(list(chain(*el))) |
|
|
|
mergedPredListListForm.append(list(chain(*el))) |
|
|
|
XClassifiers = df_cv_results_classifiers_metrics |
|
|
|
XClassifiers = preProcessResults[4] |
|
|
|
|
|
|
|
|
|
|
|
PredictionSpace = FunTsne(mergedPredListListForm) |
|
|
|
PredictionSpace = FunTsne(mergedPredListListForm) |
|
|
|
DataSpace = FunTsne(XData) |
|
|
|
DataSpace = FunTsne(XData) |
|
|
|
ModelSpace = FunMDS(XClassifiers) |
|
|
|
ModelSpaceMDS = FunMDS(XClassifiers) |
|
|
|
|
|
|
|
ModelSpaceTSNE = FunTsne(XClassifiers) |
|
|
|
|
|
|
|
ModelSpaceTSNE = ModelSpaceTSNE.tolist() |
|
|
|
global ClassifierIDsList |
|
|
|
global ClassifierIDsList |
|
|
|
key = 0 |
|
|
|
key = 0 |
|
|
|
EnsembleModel(ClassifierIDsList, key) |
|
|
|
EnsembleModel(ClassifierIDsList, key) |
|
|
|
DataSpaceList = DataSpace.tolist() |
|
|
|
|
|
|
|
PredictionSpaceList = PredictionSpace.tolist() |
|
|
|
PredictionSpaceList = PredictionSpace.tolist() |
|
|
|
|
|
|
|
DataSpaceList = DataSpace.tolist() |
|
|
|
|
|
|
|
ReturnResults(sumPerClassifier,ModelSpaceMDS,ModelSpaceTSNE,preProcessResults,DataSpaceList,PredictionSpaceList) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def ReturnResults(sumPerClassifier,ModelSpaceMDS,ModelSpaceTSNE,preProcessResults,DataSpaceList,PredictionSpaceList): |
|
|
|
global Results |
|
|
|
global Results |
|
|
|
|
|
|
|
|
|
|
|
Results = [] |
|
|
|
Results = [] |
|
|
|
|
|
|
|
FeatureImportance = preProcessResults[1] |
|
|
|
|
|
|
|
PerClassMetrics = preProcessResults[2] |
|
|
|
|
|
|
|
FeatureAccuracy = preProcessResults[3] |
|
|
|
FeatureImportance = FeatureImportance.to_json(orient='records') |
|
|
|
FeatureImportance = FeatureImportance.to_json(orient='records') |
|
|
|
PerClassMetrics = PerClassMetrics.to_json(orient='records') |
|
|
|
PerClassMetrics = PerClassMetrics.to_json(orient='records') |
|
|
|
FeatureAccuracy = FeatureAccuracy.to_json(orient='records') |
|
|
|
FeatureAccuracy = FeatureAccuracy.to_json(orient='records') |
|
|
|
DataSpaceList = DataSpace.tolist() |
|
|
|
|
|
|
|
XDataJSON = XData.columns.tolist() |
|
|
|
XDataJSON = XData.columns.tolist() |
|
|
|
Results.append(json.dumps(sumPerClassifier)) # Position: 0 |
|
|
|
Results.append(json.dumps(sumPerClassifier)) # Position: 0 |
|
|
|
Results.append(json.dumps(ModelSpace)) # Position: 1 |
|
|
|
Results.append(json.dumps(ModelSpaceMDS)) # Position: 1 |
|
|
|
Results.append(json.dumps(classifiersIDPlusParams)) # Position: 2 |
|
|
|
Results.append(json.dumps(classifiersIDPlusParams)) # Position: 2 |
|
|
|
Results.append(FeatureImportance) # Position: 3 |
|
|
|
Results.append(FeatureImportance) # Position: 3 |
|
|
|
Results.append(PerClassMetrics) # Position: 4 |
|
|
|
Results.append(PerClassMetrics) # Position: 4 |
|
|
|
@ -430,11 +489,9 @@ def InitializeEnsemble(): |
|
|
|
Results.append(json.dumps(classifiersIDwithFI)) # Position: 9 |
|
|
|
Results.append(json.dumps(classifiersIDwithFI)) # Position: 9 |
|
|
|
Results.append(json.dumps(DataSpaceList)) # Position: 10 |
|
|
|
Results.append(json.dumps(DataSpaceList)) # Position: 10 |
|
|
|
Results.append(json.dumps(PredictionSpaceList)) # Position: 11 |
|
|
|
Results.append(json.dumps(PredictionSpaceList)) # Position: 11 |
|
|
|
|
|
|
|
Results.append(json.dumps(ModelSpaceTSNE)) # Position: 12 |
|
|
|
return Results |
|
|
|
return Results |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# Retrieve data from client |
|
|
|
# Retrieve data from client |
|
|
|
@cross_origin(origin='localhost',headers=['Content-Type','Authorization']) |
|
|
|
@cross_origin(origin='localhost',headers=['Content-Type','Authorization']) |
|
|
|
@app.route('/data/ServerRequestSelPoin', methods=["GET", "POST"]) |
|
|
|
@app.route('/data/ServerRequestSelPoin', methods=["GET", "POST"]) |
|
|
|
@ -456,11 +513,12 @@ def FeatureSelPerModel(): |
|
|
|
RetrieveModelsPar = json.loads(RetrieveModelsPar) |
|
|
|
RetrieveModelsPar = json.loads(RetrieveModelsPar) |
|
|
|
RetrieveModelsParRed = [] |
|
|
|
RetrieveModelsParRed = [] |
|
|
|
print(RetrieveModelsPar['brushedAll']) # FIX THIS THING!!!!! |
|
|
|
print(RetrieveModelsPar['brushedAll']) # FIX THIS THING!!!!! |
|
|
|
for j, i in enumerate(RetrieveModelsPar['brushedAll']): |
|
|
|
#for j, i in enumerate(RetrieveModelsPar['brushedAll']): |
|
|
|
print(j) |
|
|
|
# print(j) |
|
|
|
RetrieveModelsParRed = [for j, i in enumerate(RetrieveModelsPar['brushedAll']) if j not in ClassifierIDsList] |
|
|
|
#RetrieveModelsParRed = [for j, i in enumerate(RetrieveModelsPar['brushedAll']) if j not in ClassifierIDsList] |
|
|
|
|
|
|
|
|
|
|
|
RetrieveModelsParPandas = pd.DataFrame(RetrieveModelsParRed) |
|
|
|
#RetrieveModelsParPandas = pd.DataFrame(RetrieveModelsParRed) |
|
|
|
|
|
|
|
RetrieveModelsParPandas = pd.DataFrame(RetrieveModelsPar) |
|
|
|
RetrieveModelsParPandas = RetrieveModelsParPandas.drop(columns=['performance']) |
|
|
|
RetrieveModelsParPandas = RetrieveModelsParPandas.drop(columns=['performance']) |
|
|
|
RetrieveModelsParPandas = RetrieveModelsParPandas.to_dict(orient='list') |
|
|
|
RetrieveModelsParPandas = RetrieveModelsParPandas.to_dict(orient='list') |
|
|
|
print(RetrieveModelsParPandas) |
|
|
|
print(RetrieveModelsParPandas) |
|
|
|
@ -557,7 +615,6 @@ def EnsembleModel (ClassifierIDsList, keyRetrieved): |
|
|
|
for j, each in enumerate(resultsList[index][1]): |
|
|
|
for j, each in enumerate(resultsList[index][1]): |
|
|
|
all_classifiers.append(make_pipeline(ColumnSelector(cols=columnsReduce[j]), RandomForestClassifier().set_params(**each))) |
|
|
|
all_classifiers.append(make_pipeline(ColumnSelector(cols=columnsReduce[j]), RandomForestClassifier().set_params(**each))) |
|
|
|
del columnsReduce[0:len(resultsList[index][1])] |
|
|
|
del columnsReduce[0:len(resultsList[index][1])] |
|
|
|
print(all_classifiers) |
|
|
|
|
|
|
|
sclf = StackingCVClassifier(classifiers=all_classifiers, |
|
|
|
sclf = StackingCVClassifier(classifiers=all_classifiers, |
|
|
|
use_probas=True, |
|
|
|
use_probas=True, |
|
|
|
meta_classifier=lr, |
|
|
|
meta_classifier=lr, |
|
|
|
@ -566,7 +623,6 @@ def EnsembleModel (ClassifierIDsList, keyRetrieved): |
|
|
|
else: |
|
|
|
else: |
|
|
|
for index, eachelem in enumerate(algorithmList): |
|
|
|
for index, eachelem in enumerate(algorithmList): |
|
|
|
if (eachelem == 'KNN'): |
|
|
|
if (eachelem == 'KNN'): |
|
|
|
print(resultsList[index][1]) |
|
|
|
|
|
|
|
for j, each in enumerate(resultsList[index][1]): |
|
|
|
for j, each in enumerate(resultsList[index][1]): |
|
|
|
all_classifiersSelection.append(make_pipeline(ColumnSelector(cols=columnsReduce[j]), KNeighborsClassifier().set_params(**each))) |
|
|
|
all_classifiersSelection.append(make_pipeline(ColumnSelector(cols=columnsReduce[j]), KNeighborsClassifier().set_params(**each))) |
|
|
|
del columnsReduce[0:len(resultsList[index][1])] |
|
|
|
del columnsReduce[0:len(resultsList[index][1])] |
|
|
|
@ -574,7 +630,6 @@ def EnsembleModel (ClassifierIDsList, keyRetrieved): |
|
|
|
for j, each in enumerate(resultsList[index][1]): |
|
|
|
for j, each in enumerate(resultsList[index][1]): |
|
|
|
all_classifiersSelection.append(make_pipeline(ColumnSelector(cols=columnsReduce[j]), RandomForestClassifier().set_params(**each))) |
|
|
|
all_classifiersSelection.append(make_pipeline(ColumnSelector(cols=columnsReduce[j]), RandomForestClassifier().set_params(**each))) |
|
|
|
del columnsReduce[0:len(resultsList[index][1])] |
|
|
|
del columnsReduce[0:len(resultsList[index][1])] |
|
|
|
print(all_classifiersSelection) |
|
|
|
|
|
|
|
sclf = StackingCVClassifier(classifiers=all_classifiersSelection, |
|
|
|
sclf = StackingCVClassifier(classifiers=all_classifiersSelection, |
|
|
|
use_probas=True, |
|
|
|
use_probas=True, |
|
|
|
meta_classifier=lr, |
|
|
|
meta_classifier=lr, |
|
|
|
@ -687,6 +742,7 @@ def RetrieveModelsParam(): |
|
|
|
algorithm = RetrieveModelsPar['algorithm'] |
|
|
|
algorithm = RetrieveModelsPar['algorithm'] |
|
|
|
RetrieveModelsParPandas = pd.DataFrame(RetrieveModelsPar['brushed']) |
|
|
|
RetrieveModelsParPandas = pd.DataFrame(RetrieveModelsPar['brushed']) |
|
|
|
RetrieveModelsParPandas = RetrieveModelsParPandas.drop(columns=['performance']) |
|
|
|
RetrieveModelsParPandas = RetrieveModelsParPandas.drop(columns=['performance']) |
|
|
|
|
|
|
|
RetrieveModelsParPandas = RetrieveModelsParPandas.drop(columns=['model']) |
|
|
|
RetrieveModelsParPandas = RetrieveModelsParPandas.to_dict(orient='list') |
|
|
|
RetrieveModelsParPandas = RetrieveModelsParPandas.to_dict(orient='list') |
|
|
|
RetrieveModels = {} |
|
|
|
RetrieveModels = {} |
|
|
|
for key, value in RetrieveModelsParPandas.items(): |
|
|
|
for key, value in RetrieveModelsParPandas.items(): |
|
|
|
|
